Searcharxiv⌕ Search

arXiv subjects

Xiao Zhang

Publications and source records attributed to Xiao Zhang.

At least 109 records · Page 6Linked to original sources
arXiv2026

GPG: A Simple and Strong Reinforcement Learning Baseline for Model Reasoning

Reinforcement Learning (RL) can directly enhance the reasoning capabilities of large language models without extensive reliance on Supervised Fine-Tuning (SFT). In this work, we revisit the traditional Policy Gradient (PG) mechanism and propose a minimalist RL approach termed Group Policy Gradient (GPG). Unlike conventional methods, GPG directly optimize the original RL objective, thus obviating the need for surrogate loss functions. By eliminating the critic and reference models, avoiding KL divergence constraints, and addressing the advantage and gradient estimation bias, our approach significantly simplifies the training process compared to Group Relative Policy Optimization (GRPO). Our approach achieves superior performance without relying on auxiliary techniques or adjustments. As illustrated in Figure 1, extensive experiments demonstrate that our method not only reduces computational costs but also consistently outperforms GRPO across various unimodal and multimodal tasks. Our code is available at https://github.com/AMAP-ML/GPG.

cs.LG↗
arXiv2026

DLLM-Searcher: Adapting Diffusion Large Language Model for Search Agents

Recently, Diffusion Large Language Models (dLLMs) have demonstrated unique efficiency advantages, enabled by their inherently parallel decoding mechanism and flexible generation paradigm. Meanwhile, despite the rapid advancement of Search Agents, their practical deployment is constrained by a fundamental limitation, termed as 1) Latency Challenge: the serial execution of multi-round reasoning, tool calling, and tool response waiting under the ReAct agent paradigm induces severe end-to-end latency. Intuitively, dLLMs can leverage their distinctive strengths to optimize the operational efficiency of agents under the ReAct agent paradigm. Practically, existing dLLM backbones face the 2) Agent Ability Challenge. That is, existing dLLMs exhibit remarkably weak reasoning and tool-calling capabilities, preventing these advantages from being effectively realized in practice. In this paper, we propose DLLM-Searcher, an optimization framework for dLLM-based Search Agents. To solve the Agent Ability Challenge, we design a two-stage post-training pipeline encompassing Agentic Supervised Fine-Tuning (Agentic SFT) and Agentic Variance-Reduced Preference Optimization Agentic VRPO, which enhances the backbone dLLM's information seeking and reasoning capabilities. To mitigate the Latency Challenge, we leverage the flexible generation mechanism of dLLMs and propose a novel agent paradigm termed Parallel-Reasoning and Acting P-ReAct. P-ReAct guides the model to prioritize decoding tool_call instructions, thereby allowing the model to keep thinking while waiting for the tool's return. Experimental results demonstrate that DLLM-Searcher achieves performance comparable to mainstream LLM-based search agents and P-ReAct delivers approximately 15% inference acceleration. Our code is available at https://anonymous.4open.science/r/DLLM-Searcher-553C

cs.AI↗
arXiv2026

Hexcute: A Compiler Framework for Automating Layout Synthesis in GPU Programs

Efficient GPU programming is crucial for achieving high performance in deep learning (DL) applications. The performance of GPU programs depends on how data is parallelized across threads and arranged within memory subsystems. The mapping functions describing tensors on GPUs are known as \emph{tensor layouts}. Low-level programming frameworks, such as CUTLASS and Hidet, provide expressive layout abstractions but often require \emph{considerable programming effort} to manually specify optimal layouts. High-level GPU programming languages, such as Triton, rely on compiler heuristics to generate dataflow, layouts, and pipelining strategies in GPU programs. However, the heuristics for dataflow and pipelining strategies are not generalizable to complex operators. To balance expressiveness and programmability, we propose Hexcute, a compiler framework that automates layout synthesis while providing explicit control over dataflow and pipelining. Hexcute formalizes layout synthesis as a constraint programming problem and solves it with a type-inference-based algorithm. This approach enables systematic exploration of optimal layouts and instructions. Our evaluation shows that Hexcute matches the performance of libraries like cuBLAS and FlashAttention on GEMM, Attention, and their variants, while reducing the amount of code by 1.27$\times$-7.94$\times$ compared to CUTLASS. For mixed-type mixture-of-experts (MoE) operators, Hexcute achieves an average speedup of 6.46$\times$ over Triton. In the end-to-end evaluations of vLLM, Hexcute delivers up to 2.60$\times$ speedup on DeepSeek-R1-AWQ and 2.04$\times$ on a Mamba-based model.

cs.LG↗
arXiv2026

Harder Is Better: Boosting Mathematical Reasoning via Difficulty-Aware GRPO and Multi-Aspect Question Reformulation

Reinforcement Learning with Verifiable Rewards (RLVR) offers a robust mechanism for enhancing mathematical reasoning in large models. However, we identify a systematic lack of emphasis on more challenging questions in existing methods from both algorithmic and data perspectives, despite their importance for refining underdeveloped capabilities. Algorithmically, widely used Group Relative Policy Optimization (GRPO) suffers from an implicit imbalance where the magnitude of policy updates is lower for harder questions. Data-wise, augmentation approaches primarily rephrase questions to enhance diversity without systematically increasing intrinsic difficulty. To address these issues, we propose a two-dual MathForge framework to improve mathematical reasoning by targeting harder questions from both perspectives, which comprises a Difficulty-Aware Group Policy Optimization (DGPO) algorithm and a Multi-Aspect Question Reformulation (MQR) strategy. Specifically, DGPO first rectifies the implicit imbalance in GRPO via difficulty-balanced group advantage estimation, and further prioritizes harder questions by difficulty-aware question-level weighting. Meanwhile, MQR reformulates questions across multiple aspects to increase difficulty while maintaining the original gold answer. Overall, MathForge forms a synergistic loop: MQR expands the data frontier, and DGPO effectively learns from the augmented data. Extensive experiments show that MathForge significantly outperforms existing methods on various mathematical reasoning tasks. The code and augmented data are all available at https://github.com/AMAP-ML/MathForge.

cs.AI↗
arXiv2026

Pressure-induced superconductivity in topological insulator Ge2Bi2Te5 and the evolution with Mn doping

Introducing superconductivity (SC) or magnetism into topological insulators (TIs) can give rise to novel quantum states and exotic physical phenomena. Here, we report a high-pressure transport study on the TI Ge2Bi2Te5 and its Mn-doped counterparts. The application of pressure induces a SC in Ge2Bi2Te5, which shows a dome-shape phase diagram with the maximum Tc of 7.6 K at 23 GPa. Doping Mn into Ge2Bi2Te5 introduces an antiferromagnetic order at ambient pressure and strongly weakens the pressure-induced SC, demonstrating that magnetism and SC compete in this material system. Present study provides a new platform for investigating the interplay among band topology, magnetism, and SC.

cond-mat.supr-con↗
arXiv2026

Superconductivity in W3Re2C with chiral structure

We discover superconductivity in cubic W3Re2C with chiral structure and the superconducting transition temperature Tc is about 6.2 K. Detailed characterizations and analysis indicate that W3Re2C is a bulk type-II BCS superconductor with full isotropic gap. Moreover, first-principles calculations indicate that the electron-phonon coupling primarily arises from interactions between W/Re 5d electronic states and their low-frequency phonons. Furthermore, the breaking of inversion symmetry in W3Re2C facilitates the emergence of Weyl points in the electronic structure. Therefore, W3Re2C can serve as a promising platform for investigating the influences of chiral structure on both superconductivity and band topology.

cond-mat.supr-con↗
arXiv2026

Process In-Context Learning: Enhancing Mathematical Reasoning via Dynamic Demonstration Insertion

In-context learning (ICL) has proven highly effective across diverse large language model (LLM) tasks. However, its potential for enhancing tasks that demand step-by-step logical deduction, such as mathematical reasoning, remains underexplored. A core limitation of existing ICL approaches is their static use of demonstrations: examples are pre-selected before inference and remain fixed, failing to adapt to the dynamic confusion points that often arise during multi-step reasoning such as ambiguous calculations or logical gaps. These unresolved confusion points can lead to cascading errors that degrade final accuracy. To tackle this issue, we propose Process In-Context Learning (PICL), a dynamic demonstration integration framework designed to boost mathematical reasoning by responding to real-time inference needs. PICL operates in two stages: 1)~it identifies potential confusion points by analyzing semantics and entropy in the reasoning process and summarizes their core characteristics; 2)~upon encountering these points, it retrieves relevant demonstrations from the demonstration pool that match the confusion context and inserts them directly into the ongoing reasoning process to guide subsequent steps. Experiments show that PICL outperforms baseline methods by mitigating mid-inference confusion, highlighting the value of adaptive demonstration insertion in complex mathematical reasoning.

cs.AI↗
arXiv2026

Thermodynamic and electronic properties of rutile Sn$_{1-x}$Ge$_x$O$_2$ alloys from first principles

Rutile Sn$_{1-x}$Ge$_x$O$_{2}$ alloys are promising materials for high-power electronic applications due to their dopability and tunable ultra-wide band gaps. We use first-principles density functional theory and statistical mechanics to investigate the crystallographic, electronic, and thermodynamic properties of rutile $\text{Sn}_{1-x}\text{Ge}_x\text{O}_2$ alloys. We predict that the lattice parameters follow Vegard's law, while band gaps calculated with the hybrid HSE06 functional exhibit strong bowing, consistent with experiment. We also predict that the disordered phase has a large positive mixing enthalpy and a slight tendency for Ge-Sn clustering, indicated by weakly negative short-range order parameters. This large positive mixing enthalpy produces a miscibility gap with a critical temperature above 2300 K, implying that the high Ge and Sn solubilities observed in thin-film synthesis cannot be explained by the incoherent phase diagram alone. We demonstrate that coherency strain during epitaxial growth substantially alters phase stability. Calculations of the coherent spinodal show significant suppression of the miscibility gap, reducing the critical temperature to $\approx 900$ K. These coherent phase boundaries account for the experimentally observed high solubilities at typical growth temperatures. Our results indicate that coherency strain stabilizes these metastable alloys and enables bandgap engineering in this ultrawide-bandgap material system.

cond-mat.mtrl-sci↗
arXiv2026

When Personalization Misleads: Understanding and Mitigating Hallucinations in Personalized LLMs

Personalized large language models (LLMs) adapt model behavior to individual users to enhance user satisfaction, yet personalization can inadvertently distort factual reasoning. We show that when personalized LLMs face factual queries, there exists a phenomenon where the model generates answers aligned with a user's prior history rather than the objective truth, resulting in personalization-induced hallucinations that degrade factual reliability and may propagate incorrect beliefs, due to representational entanglement between personalization and factual representations. To address this issue, we propose Factuality-Preserving Personalized Steering (FPPS), a lightweight inference-time approach that mitigates personalization-induced factual distortions while preserving personalized behavior. We further introduce PFQABench, the first benchmark designed to jointly evaluate factual and personalized question answering under personalization. Experiments across multiple LLM backbones and personalization methods show that FPPS substantially improves factual accuracy while maintaining personalized performance.

cs.CL↗
arXiv2026

Semi-Tensor-Product Based Convolutional Neural Networks

The semi-tensor product of vectors generalizes the conventional inner product, enabling algebraic operations between vectors of different dimensions. Building upon this foundation, we introduce a domain-based convolutional product and integrate it with the STP to formulate a padding-free convolutional operation. This new operation inherently avoids zero or other artificial padding, thereby eliminating redundant information and boundary artifacts commonly present in conventional convolutional neural networks. Based on this operation, we further develop an STP-based CNN framework that extends convolutional computation to irregular and cross-dimensional data domains. Applications to image processing and third-order signal identification demonstrate the proposed method's effectiveness in handling irregular, incomplete, and high-dimensional data without the distortions caused by padding.

eess.SY↗
arXiv2026

Towards Compositional Generalization in LLMs for Smart Contract Security: A Case Study on Reentrancy Vulnerabilities

Large language models (LLMs) demonstrate remarkable capabilities in natural language understanding and generation. Despite being trained on large-scale, high-quality data, LLMs still fail to outperform traditional static analysis tools in specialized domains like smart contract vulnerability detection. To address this issue, this paper proposes a post-training algorithm based on atomic task decomposition and fusion. This algorithm aims to achieve combinatorial generalization under limited data by decomposing complex reasoning tasks. Specifically, we decompose the reentrancy vulnerability detection task into four linearly independent atomic tasks: identifying external calls, identifying state updates, identifying data dependencies between external calls and state updates, and determining their data flow order. These tasks form the core components of our approach. By training on synthetic datasets, we generate three compiler-verified datasets. We then employ the Slither tool to extract structural information from the control flow graph and data flow graph, which is used to fine-tune the LLM's adapter. Experimental results demonstrate that low-rank normalization fusion with the LoRA adapter improves the LLM's reentrancy vulnerability detection accuracy to 98.2%, surpassing state-of-the-art methods. On 31 real-world contracts, the algorithm achieves a 20% higher recall than traditional analysis tools.

cs.CR↗
arXiv2026

Curator: Efficient Vector Search with Low-Selectivity Filters

Embedding-based dense retrieval has become the cornerstone of many critical applications, where approximate nearest neighbor search (ANNS) queries are often combined with filters on labels such as dates and price ranges. Graph-based indexes achieve state-of-the-art performance on unfiltered ANNS but encounter connectivity breakdown on low-selectivity filtered queries, where qualifying vectors become sparse and the graph structure among them fragments. Recent research proposes specialized graph indexes that address this issue by expanding graph degree, which incurs prohibitively high construction costs. Given these inherent limitations of graph-based methods, we argue for a dual-index architecture and present Curator, a partition-based index that complements existing graph-based approaches for low-selectivity filtered ANNS. Curator builds specialized indexes for different labels within a shared clustering tree, where each index adapts to the distribution of its qualifying vectors to ensure efficient search while sharing structure to minimize memory overhead. The system also supports incremental updates and handles arbitrary complex predicates beyond single-label filters by efficiently constructing temporary indexes on the fly. Our evaluation demonstrates that integrating Curator with state-of-the-art graph indexes reduces low-selectivity query latency by up to 20.9x compared to pre-filtering fallback, while increasing construction time and memory footprint by only 5.5% and 4.3%, respectively.

cs.DB↗
arXiv2026

GR-Dexter Technical Report

Vision-language-action (VLA) models have enabled language-conditioned, long-horizon robot manipulation, but most existing systems are limited to grippers. Scaling VLA policies to bimanual robots with high degree-of-freedom (DoF) dexterous hands remains challenging due to the expanded action space, frequent hand-object occlusions, and the cost of collecting real-robot data. We present GR-Dexter, a holistic hardware-model-data framework for VLA-based generalist manipulation on a bimanual dexterous-hand robot. Our approach combines the design of a compact 21-DoF robotic hand, an intuitive bimanual teleoperation system for real-robot data collection, and a training recipe that leverages teleoperated robot trajectories together with large-scale vision-language and carefully curated cross-embodiment datasets. Across real-world evaluations spanning long-horizon everyday manipulation and generalizable pick-and-place, GR-Dexter achieves strong in-domain performance and improved robustness to unseen objects and unseen instructions. We hope GR-Dexter serves as a practical step toward generalist dexterous-hand robotic manipulation.

cs.RO↗
arXiv2026

AeroSketch: Near-Optimal Time Matrix Sketch Framework for Persistent, Sliding Window, and Distributed Streams

Many real-world matrix datasets arrive as high-throughput vector streams, making it impractical to store or process them in their entirety. To enable real-time analytics under limited computational, memory, and communication resources, matrix sketching techniques have been developed over recent decades to provide compact approximations of such streaming data. Some algorithms have achieved optimal space and communication complexity. However, these approaches often require frequent time-consuming matrix factorization operations. In particular, under tight approximation error bounds, each matrix factorization computation incurs cubic time complexity, thereby limiting their update efficiency. In this paper, we introduce AeroSketch, a novel matrix sketching framework that leverages recent advances in randomized numerical linear algebra (RandNLA). AeroSketch achieves optimal communication and space costs while delivering near-optimal update time complexity (within logarithmic factors) across persistent, sliding window, and distributed streaming scenarios. Extensive experiments on both synthetic and real-world datasets demonstrate that AeroSketch consistently outperforms state-of-the-art methods in update throughput. In particular, under tight approximation error constraints, AeroSketch reduces the cubic time complexity to the quadratic level. Meanwhile, it maintains comparable approximation quality while retaining optimal communication and space costs.

cs.DB↗
arXiv2026

Cooperative concurrence of 4f and 3d flat bands in kagome heavy-fermion metal YbCr6Ge6

Flat-band (FB) systems originating from special lattice geometry like in kagome metals as well as localized orbitals in the materials such as heavy-fermion (HF) compounds have induced intensive interest due to their band topology and strong electron correlation effects, leading to emergent quantum states of matter. However, the question of how these two distinct FBs coexist and interact remains unsettled. Here, we report that YbCr6Ge6 hosting both Cr-kagome lattice and Yb-4f electrons exhibits HF behaviors and a robust antiferromagnetic ground state with transition temperature TN = 3 K, significantly higher than other similar kagome metals with Yb ions. Angle-resolved photoemission spectroscopy measurements reveal the coexistence of FBs originating from both Cr-kagome lattice and localized Yb-4f electrons near Fermi energy level EF. More importantly, the clear spectroscopic signatures of a hybridization of Yb-4f FB with kagome-lattice-derived conduction bands and the high density of states of Cr-kagome FB near EF provide the underlying microscopic mechanisms of HF behaviors and enhanced antiferromagnetism in YbCr6Ge6. Our findings demonstrate that the novel kagome HF metals can not only host the cooperative coexistence of two different types of FBs, but also provide a paradigm material platform to explore the exotic correlated topological quantum phenomena.

cond-mat.str-el↗
arXiv2026

Revisiting Chain-of-Thought Prompting: Zero-shot Can Be Stronger than Few-shot

In-Context Learning (ICL) is an essential emergent ability of Large Language Models (LLMs), and recent studies introduce Chain-of-Thought (CoT) to exemplars of ICL to enhance the reasoning capability, especially in mathematics tasks. However, given the continuous advancement of model capabilities, it remains unclear whether CoT exemplars still benefit recent, stronger models in such tasks. Through systematic experiments, we find that for recent strong models such as the Qwen2.5 series, adding traditional CoT exemplars does not improve reasoning performance compared to Zero-Shot CoT. Instead, their primary function is to align the output format with human expectations. We further investigate the effectiveness of enhanced CoT exemplars, constructed using answers from advanced models such as \texttt{Qwen2.5-Max} and \texttt{DeepSeek-R1}. Experimental results indicate that these enhanced exemplars still fail to improve the model's reasoning performance. Further analysis reveals that models tend to ignore the exemplars and focus primarily on the instructions, leading to no observable gain in reasoning ability. Overall, our findings highlight the limitations of the current ICL+CoT framework in mathematical reasoning, calling for a re-examination of the ICL paradigm and the definition of exemplars.

cs.CL↗
arXiv2026

Sandwich Reasoning: An Answer-Reasoning-Answer Approach for Low-Latency Query Correction

Query correction is a critical entry point in modern search pipelines, demanding high accuracy strictly within real-time latency constraints. Chain-of-Thought (CoT) reasoning improves accuracy but incurs prohibitive latency for real-time query correction. A potential solution is to output an answer before reasoning to reduce latency; however, under autoregressive decoding, the early answer is independent of subsequent reasoning, preventing the model from leveraging its reasoning capability to improve accuracy. To address this issue, we propose Sandwich Reasoning (SandwichR), a novel approach that explicitly aligns a fast initial answer with post-hoc reasoning, enabling low-latency query correction without sacrificing reasoning-aware accuracy. SandwichR follows an Answer-Reasoning-Answer paradigm, producing an initial correction, an explicit reasoning process, and a final refined correction. To align the initial answer with post-reasoning insights, we design a consistency-aware reinforcement learning (RL) strategy: a dedicated consistency reward enforces alignment between the initial and final corrections, while margin-based rejection sampling prioritizes borderline samples where reasoning drives the most impactful corrective gains. Additionally, we construct a high-quality query correction dataset, addressing the lack of specialized benchmarks for complex query correction. Experimental results demonstrate that SandwichR achieves SOTA accuracy comparable to standard CoT while delivering a 40-70% latency reduction, resolving the latency-accuracy trade-off in online search.

cs.AI↗
arXiv2026

Nematic-fluctuation-mediated superconductivity in CuxTiSe2

The interplay among electronic nematicity, charge density wave, and superconductivity in correlated electronic systems has induced extensive research interest. Here, we discover the existence of nematic fluctuations in TiSe2 single crystal and investigate its evolution with Cu intercalation. It is observed that the elastoresistivity coefficient mEg exhibits a divergent temperature dependence following a Curie-Weiss law at high temperature. Upon Cu intercalation, the characteristic temperature T* of nematic fluctuation is progressively suppressed and becomes near zero when the superconductivity is optimized. Further intercalation of Cu leads to the sign change of T* and the suppression of superconductivity. These results strongly indicate that nematic phase transition may play a vital role in enhancing superconductivity in CuxTiSe2. Therefore, CuxTiSe2 provides a unique material platform to explore the nematic-fluctuation-mediated superconductivity.

cond-mat.supr-con↗