Search arXiv⌕ Search

arXiv subjects

Jian Shao

Publications and source records attributed to Jian Shao.

At least 19 recordsLinked to original sources

RuACT: Recovering Inactive Rubrics as Adaptive Critics and Teachers

Rubric-based reinforcement learning (RL) extends RL with verifiable rewards to open-ended tasks by scoring each response against query-specific rubrics. Under group-relative optimization, however, a rubric that every sampled response fails (unreachable) or passes (saturated) gives all responses the same outcome and no learning signal; in Qwen3-4B medical training, 60.88% of query-rubric observations are inactive. We find that the two states lack different supervision: stating an unreachable rubric in the prompt raises its satisfaction from 1.13% to 91.25%, whereas responses that all pass a saturated rubric still differ in quality. We therefore propose RuACT (Rubrics as Adaptive Critics and Teachers), which assigns each inactive rubric the role that supplies what it lacks. Rubric as Teacher uses unreachable rubrics as privileged information for reward-gated on-policy self-distillation, and Rubric as Critic decomposes saturated rubrics into child rubrics that separate current responses while sharing the parent's weight. RuACT outperforms GRPO, rubric reweighting, and rubric-guided exploration on all four medical and scientific benchmarks with Qwen3-4B and achieves the best average with Qwen3-1.7B. On initially unreachable rubrics, unguided recovery rises from 35.02% to 42.86%, while child refinement improves reward-quality alignment on independent rollouts from 57.74% to 60.77%. Code is available at https://github.com/ZJU-REAL/RuACT.

cs.CL↗

SWE-Ladder: Learning to Localize, Reproduce, and Patch via Progressive Privileged Self-Distillation

Resolving a repository-level issue typically requires a software-engineering (SWE) agent to localize the faulty code, reproduce the failure, and patch the code. Reinforcement learning trains such agents from a single test result on the final patch, while on-policy self-distillation (OPSD) adds token-level supervision from a teacher with privileged evidence, exposing the full reference patch at every step. Neither provides stage-specific guidance for localization or reproduction. We propose SWE-Ladder, which turns the verified repair of each training instance into stage-specific evidence: the changed files and lines for localization, the fail-to-pass tests for reproduction, and the diff for patching. A rule-based router assigns each step of an end-to-end rollout to a stage based on its tool call, a frozen teacher sees only the matching evidence, and the stages are distilled in repair order while GRPO optimizes the full trajectory. With Qwen3-4B-Instruct-2507 on SWE-bench Verified under the R2E-Gym scaffold, SWE-Ladder improves pass@1 from 9.6% to 16.4%, compared with 10.4% for GRPO and 12.4% for OPSD with the full patch. Under the OpenHands scaffold, it improves pass@1 from 13.0% to 17.6%. Reversing the stage order under R2E-Gym reduces pass@1 to 11.4%.

cs.AI↗

Scattering-Mediated NLOS Ultraviolet Communication: Channel Characterization, Modeling, and Data Transmission

Solar-blind ultraviolet (UV) optical wireless communication (OWC) enables non-line-of-sight (NLOS) transmission through atmospheric scattering. This work presents a comprehensive experimental and simulation study of a 100 m outdoor UVC link with comparative measurements under LOS-blocked and open-field conditions. The results show that LOS-blocked scattering-mediated links exhibit substantially higher path loss and narrower bandwidth than corresponding open-field measurements, highlighting the impact of residual LOS components in practical diffused beam UVC experiments. A photon-tracing Monte Carlo (MC) framework incorporating realistic transmitter/receiver characteristics and obstacle geometry is further developed to reproduce the measured channel characteristics with high accuracy. Despite the severe path loss and pulse broadening effects, reliable 20-200 kbps transmission is experimentally demonstrated over the 100 m LOS-blocked link using on-off keying (OOK) and pulse-position modulation (PPM), representing the longest-distance and highest-data-rate experimental demonstration of an explicitly LOS-blocked scattering-mediated UVC link reported to date.

eess.SP↗

Solar-Blind Ultraviolet Communication: Taxonomy and Experimental Characterization

Solar-blind ultraviolet communication (UVC) exploits near-zero solar background radiation and strong atmospheric scattering in the 200-280 nm band, enabling photon-level daytime communication and propagation modes beyond conventional line-of-sight (LOS) links. However, existing studies mainly focus on individual channel types, and a unified channel classification is still lacking. This paper presents a taxonomy of all four fundamental solar-blind UVC channels together with representative experimental validations and a comparative performance analysis. The proposed framework provides practical guidelines for selecting appropriate channel architectures for future solar-blind UVC systems.

eess.SP↗

IterSynth: Rethinking Deep Search Agents via Role-Decoupled Iterative Synthesis

Deep search requires LLM agents to decompose complex queries, search for evidence, and synthesize grounded answers, yet existing ReAct-style agents suffer from two limitations: role coupling, where one policy must handle planning, evidence use, and synthesis; and context accumulation, where growing search histories introduce noise and obscure useful information. To address these issues, we propose IterSynth, a role-decoupled and summary-based paradigm that alternates between a Planner for identifying information needs and a Synthesizer for integrating evidence into an evolving summary state. This design separates planning from synthesis while using the summary as the persistent state of search, reducing both capability coupling and context noise. To train IterSynth effectively, we further introduce Role-Decoupled Policy Optimization (RDPO) for reinforcement learning, which combines terminal outcome rewards with turn-level rubric evaluations and computes role-specific advantages for more precise credit assignment. Experiments on five long-horizon deep-search benchmarks such as BrowseComp and Xbench-DS show that IterSynth-8B achieves an average score of 50.7, surpassing the strongest prior $\leq$8B agent by +4.2\%. Moreover, IterSynth serves as a model-agnostic prompting paradigm, delivering substantial zero-shot gains over ReAct and similar prompting paradigms on frontier proprietary models.

cs.CL↗

Electric-current control of anomalous Hall effect

We demonstrate robust and reversible electric-current control of the anomalous Hall effect (AHE) in a two-dimensional WTe2/Fe3GeTe2 (FGT) stack. Applying a current through Td-WTe2 leads to a giant modulation of the AHE of the adjacent FGT layer, with the relative change of the AHE conductivity exceeding 180%. Control experiments show that i) the observed effect is absent in pure FGT, ii) the modulation weakens in thicker FGT films, confirming its interfacial origin, and iii) the modulation peaks for bilayer WTe2, indicating that the Berry-curvature dipole (BCD) plays the dominant role in the modulation. We propose that the charge current I generates an out-of-plane magnetization Mz via BCD in WTe2 and Mz modifies the exchange splitting of FGT via the inverse magnetic proximity effect, thereby altering its Berry curvature and nontrivially influencing the AHE. The demonstrated method of AHE control offers new possibilities for magnetism control, i.e., for the study of AHE-transistors as well as electric-current control of quantum magnets, especially magnetic insulators.

cond-mat.mes-hall↗

InftyThink+: Effective and Efficient Infinite-Horizon Reasoning via Reinforcement Learning

Large reasoning models achieve strong performance by scaling inference-time chain-of-thought, but this paradigm suffers from quadratic cost, context length limits, and degraded reasoning due to lost-in-the-middle effects. Iterative reasoning mitigates these issues by periodically summarizing intermediate thoughts, yet existing methods rely on supervised learning or fixed heuristics and fail to optimize when to summarize, what to preserve, and how to resume reasoning. We propose InftyThink+, an end-to-end reinforcement learning framework that optimizes the entire iterative reasoning trajectory, building on model-controlled iteration boundaries and explicit summarization. InftyThink+ adopts a two-stage training scheme with supervised cold-start followed by trajectory-level reinforcement learning, enabling the model to learn strategic summarization and continuation decisions. Experiments on DeepSeek-R1-Distill-Qwen-1.5B show that InftyThink+ improves accuracy by 21% on AIME24 and outperforms conventional long chain-of-thought reinforcement learning by a clear margin, while also generalizing better to out-of-distribution benchmarks. Moreover, InftyThink+ significantly reduces inference latency and accelerates reinforcement learning training, demonstrating improved reasoning efficiency alongside stronger performance.

cs.CL↗

Nitrogen doping induced metal-insulator transition with iso-symmetric character in rutile VO2

Metal-insulator transitions (MITs) in correlated oxides offer immense potential for next-generation Mottronic devices. However, their integration into practical applications is often hindered by the coupling of MITs with symmetry-lowering structural phase transitions, which limits switching speed and endurance. In this study, we engineered an iso-symmetric MIT on average in epitaxial rutile VO2 thin films via an in-situ nitrogen doping strategy. Nitrogen incorporation effectively suppresses V-V dimerization, enabling an iso-symmetric MIT, while preserving the original crystal symmetry. Furthermore, in-operando time-resolved optical reflectivity measurements revealed a shortened switching time in nitrogen-doped films, highlighting their enhanced performance. Our findings provide critical insights into the underlying mechanisms of MITs and introduce anion doping as a powerful tool for tailoring phase transitions in strongly correlated electron systems. This approach opens new avenues for the development of high-performance electronic and photonic devices.

cond-mat.mtrl-sci↗

Current-tunable room temperature ferromagnetism and current-driven phase transitions

It is generally assumed that the application of a charge-current in ferromagnetic metals suppresses their ferromagnetic order through trivial Joule heating. Here, we demonstrate that a charge current can instead enhance magnetic ordering. Using a WTe2/Fe3Ge2Te (FGT) stack as a model system, we show that a charge current flowing in WTe2 controls the ferromagnetic properties and magnetic phase transition of the adjacent FGT via a current-induced effective magnetic-field arising from orbital magnetization. Remarkably, the charge current drives a substantial enhancement of the Curie temperature, boosting it well above room temperature. Furthermore, we show that the charge-current enables controlled tuning of the phase transitions in FGT, which confirms the scaling behaviour of a ferromagnet-paramagnet phase transition. This work provides a pathway for integrating two-dimensional ferromagnets into spintronic functionalities at technologically relevant temperatures and for exploring novel current-driven phenomena in ferromagnetic systems.

cond-mat.mtrl-sci↗

MathFimer: Enhancing Mathematical Reasoning by Expanding Reasoning Steps through Fill-in-the-Middle Task

Mathematical reasoning represents a critical frontier in advancing large language models (LLMs). While step-by-step approaches have emerged as the dominant paradigm for mathematical problem-solving in LLMs, the quality of reasoning steps in training data fundamentally constrains the performance of the models. Recent studies have demonstrated that more detailed intermediate steps can enhance model performance, yet existing methods for step expansion either require more powerful external models or incur substantial computational costs. In this paper, we introduce MathFimer, a novel framework for mathematical reasoning step expansion inspired by the ''Fill-in-the-middle'' task from code reasoning. By decomposing solution chains into prefix-suffix pairs and training models to reconstruct missing intermediate steps, we develop a specialized model, MathFimer-7B, on our carefully curated NuminaMath-FIM dataset. We then apply these models to enhance existing mathematical reasoning datasets by inserting detailed intermediate steps into their solution chains, creating MathFimer-expanded versions. Through comprehensive experiments on multiple mathematical reasoning datasets, including MathInstruct, MetaMathQA and etc., we demonstrate that models trained on MathFimer-expanded data consistently outperform their counterparts trained on original data across various benchmarks such as GSM8K and MATH. Our approach offers a practical, scalable solution for enhancing mathematical reasoning capabilities in LLMs without relying on powerful external models or expensive inference procedures.

cs.CL↗

InftyThink: Breaking the Length Limits of Long-Context Reasoning in Large Language Models

Advanced reasoning in large language models has achieved remarkable performance on challenging tasks, but the prevailing long-context reasoning paradigm faces critical limitations: quadratic computational scaling with sequence length, reasoning constrained by maximum context boundaries, and performance degradation beyond pre-training context windows. Existing approaches primarily compress reasoning chains without addressing the fundamental scaling problem. To overcome these challenges, we introduce InftyThink, a paradigm that transforms monolithic reasoning into an iterative process with intermediate summarization. By interleaving short reasoning segments with concise progress summaries, our approach enables unbounded reasoning depth while maintaining bounded computational costs. This creates a characteristic sawtooth memory pattern that significantly reduces computational complexity compared to traditional approaches. Furthermore, we develop a methodology for reconstructing long-context reasoning datasets into our iterative format, transforming OpenR1-Math into 333K training instances. Experiments across multiple model architectures demonstrate that our approach reduces computational costs while improving performance, with Qwen2.5-Math-7B showing 3-11% improvements across MATH500, AIME24, and GPQA_diamond benchmarks. Our work challenges the assumed trade-off between reasoning depth and computational efficiency, providing a more scalable approach to complex reasoning without architectural modifications.

cs.CL↗

VerifyBench: Benchmarking Reference-based Reward Systems for Large Language Models

Large reasoning models such as OpenAI o1 and DeepSeek-R1 have demonstrated remarkable performance in complex reasoning tasks. A critical component of their training is the incorporation of reference-based reward systems within reinforcement learning (RL), where model outputs are evaluated against ground truth references. However, existing reward benchmarks focus on preference comparisons between responses rather than evaluating verification against ground truth references, leaving a critical gap in our ability to evaluate verification systems used in reasoning model training. In this paper, we introduce VerifyBench and its challenging variant VerifyBench-Hard, two benchmarks specifically designed to assess reference-based reward systems. These benchmarks are constructed through meticulous data collection and curation, followed by careful human annotation to ensure high quality. Our comprehensive evaluation reveals that while larger model-based verifiers show promise on standard cases, all current systems demonstrate substantial room for improvement on challenging instances. Through systematic analysis of performance patterns across reasoning tasks and error categories, we provide insights for advancing reference-based reward systems. These benchmarks establish a standardized framework for improving verification accuracy, ultimately enhancing reasoning capabilities in models trained via RL.

cs.CL↗

Let LRMs Break Free from Overthinking via Self-Braking Tuning

Large reasoning models (LRMs), such as OpenAI o1 and DeepSeek-R1, have significantly enhanced their reasoning capabilities by generating longer chains of thought, demonstrating outstanding performance across a variety of tasks. However, this performance gain comes at the cost of a substantial increase in redundant reasoning during the generation process, leading to high computational overhead and exacerbating the issue of overthinking. Although numerous existing approaches aim to address the problem of overthinking, they often rely on external interventions. In this paper, we propose a novel framework, Self-Braking Tuning (SBT), which tackles overthinking from the perspective of allowing the model to regulate its own reasoning process, thus eliminating the reliance on external control mechanisms. We construct a set of overthinking identification metrics based on standard answers and design a systematic method to detect redundant reasoning. This method accurately identifies unnecessary steps within the reasoning trajectory and generates training signals for learning self-regulation behaviors. Building on this foundation, we develop a complete strategy for constructing data with adaptive reasoning lengths and introduce an innovative braking prompt mechanism that enables the model to naturally learn when to terminate reasoning at an appropriate point. Experiments across mathematical benchmarks (AIME, AMC, MATH500, GSM8K) demonstrate that our method reduces token consumption by up to 60% while maintaining comparable accuracy to unconstrained models.

cs.CL↗

LAPO: Internalizing Reasoning Efficiency via Length-Adaptive Policy Optimization

Large reasoning models have achieved remarkable performance through extended chain-of-thought sequences, yet this computational freedom leads to excessive token generation even for simple problems. We present Length-Adaptive Policy Optimization (LAPO), a novel framework that transforms reasoning length control from an external constraint into an intrinsic model capability. Unlike existing approaches that impose rigid limits or rely on post-hoc interventions, LAPO enables models to internalize an understanding of appropriate reasoning depth through a two-stage reinforcement learning process. In the first stage, models learn natural reasoning patterns by discovering the statistical distribution of successful solution lengths. The second stage leverages these patterns as meta-cognitive guidance, embedding them directly within the model's reasoning context to ensure inference-time flexibility. Experiments on mathematical reasoning benchmarks demonstrate that LAPO reduces token usage by up to 40.9% while improving accuracy by 2.3%. Our analysis reveals that models trained with LAPO develop emergent abilities to allocate computational resources based on problem complexity, achieving efficient reasoning without sacrificing quality.

cs.AI↗

S^3cMath: Spontaneous Step-level Self-correction Makes Large Language Models Better Mathematical Reasoners

Self-correction is a novel method that can stimulate the potential reasoning abilities of large language models (LLMs). It involves detecting and correcting errors during the inference process when LLMs solve reasoning problems. However, recent works do not regard self-correction as a spontaneous and intrinsic capability of LLMs. Instead, such correction is achieved through post-hoc generation, external knowledge introduction, multi-model collaboration, and similar techniques. In this paper, we propose a series of mathematical LLMs called S$^3$c-Math, which are able to perform Spontaneous Step-level Self-correction for Mathematical reasoning. This capability helps LLMs to recognize whether their ongoing inference tends to contain errors and simultaneously correct these errors to produce a more reliable response. We proposed a method, which employs a step-level sampling approach to construct step-wise self-correction data for achieving such ability. Additionally, we implement a training strategy that uses above constructed data to equip LLMs with spontaneous step-level self-correction capacities. Our data and methods have been demonstrated to be effective across various foundation LLMs, consistently showing significant progress in evaluations on GSM8K, MATH, and other mathematical benchmarks. To the best of our knowledge, we are the first to introduce the spontaneous step-level self-correction ability of LLMs in mathematical reasoning.

cs.CL↗

A Survey on (M)LLM-Based GUI Agents

Graphical User Interface (GUI) Agents have emerged as a transformative paradigm in human-computer interaction, evolving from rule-based automation scripts to sophisticated AI-driven systems capable of understanding and executing complex interface operations. This survey provides a comprehensive examination of the rapidly advancing field of LLM-based GUI Agents, systematically analyzing their architectural foundations, technical components, and evaluation methodologies. We identify and analyze four fundamental components that constitute modern GUI Agents: (1) perception systems that integrate text-based parsing with multimodal understanding for comprehensive interface comprehension; (2) exploration mechanisms that construct and maintain knowledge bases through internal modeling, historical experience, and external information retrieval; (3) planning frameworks that leverage advanced reasoning methodologies for task decomposition and execution; and (4) interaction systems that manage action generation with robust safety controls. Through rigorous analysis of these components, we reveal how recent advances in large language models and multimodal learning have revolutionized GUI automation across desktop, mobile, and web platforms. We critically examine current evaluation frameworks, highlighting methodological limitations in existing benchmarks while proposing directions for standardization. This survey also identifies key technical challenges, including accurate element localization, effective knowledge retrieval, long-horizon planning, and safety-aware execution control, while outlining promising research directions for enhancing GUI Agents' capabilities. Our systematic review provides researchers and practitioners with a thorough understanding of the field's current state and offers insights into future developments in intelligent interface automation.

cs.HC↗

Structural-Temporal Coupling Anomaly Detection with Dynamic Graph Transformer

Detecting anomalous edges in dynamic graphs is an important task in many applications over evolving triple-based data, such as social networks, transaction management, and epidemiology. A major challenge with this task is the absence of structural-temporal coupling information, which decreases the ability of the representation to distinguish anomalies from normal instances. Existing methods focus on handling independent structural and temporal features with embedding models, which ignore the deep interaction between these two types of information. In this paper, we propose a structural-temporal coupling anomaly detection architecture with a dynamic graph transformer model. Specifically, we introduce structural and temporal features from two integration levels to provide anomaly-aware graph evolutionary patterns. Then, a dynamic graph transformer enhanced by two-dimensional positional encoding is implemented to capture both discrimination and contextual consistency signals. Extensive experiments on six datasets demonstrate that our method outperforms current state-of-the-art models. Finally, a case study illustrates the strength of our method when applied to a real-world task.

cs.LG↗

ProSwitch: Knowledge-Guided Instruction Tuning to Switch Between Professional and Non-Professional Responses

Large Language Models (LLMs) have demonstrated efficacy in various linguistic applications, including question answering and controlled text generation. However, studies into their ability to switch between opposite styles of responses in professional domains remain underexplored. This study introduces a novel approach, named ProSwitch, which enables a language model to switch between professional and non-professional answers, by tuning and evaluating through the guidance of domain and style knowledge. ProSwitch unfolds in three phases: LLM-augmented preparation to collect domain knowledge and QA pairs, instruction tuning to optimize LLMs with multiple levels of knowledge, and comprehensive evaluation to assess both style discrimination and reference-based quality of the generated text. Comparative analysis of ProSwitch against general and specialized LLMs reveals that our approach outperforms baselines in switching between professional and non-professional responses.

cs.CL↗