arXiv subjects

Xin Xu

Publications and source records attributed to Xin Xu.

At least 91 records · Page 5Linked to original sources

arXiv2026

ORBIT: On-policy Exploration-Exploitation for Controllable Multi-Budget Reasoning

Recent Large Reasoning Models (LRMs) achieve strong performance by leveraging long-form Chain-of-Thought (CoT) reasoning, but uniformly applying overlong reasoning at inference time incurs substantial and often unnecessary computational cost. To address this, prior work explores various strategies to infer an appropriate reasoning budget from the input. However, such approaches are unreliable in the worst case, as estimating the minimal required reasoning effort is fundamentally difficult, and they implicitly fix the trade-off between reasoning cost and accuracy during training, limiting flexibility under varying deployment scenarios. Motivated by these limitations, we propose ORBIT, a controllable multi-budget reasoning framework with well-separated reasoning modes triggered by input. ORBIT employs multi-stage reinforcement learning to discover Pareto-optimal reasoning behaviors at each effort, followed by on-policy distillation to fuse these behaviors into a single unified model. Experiments show that ORBIT achieves (1) controllable reasoning behavior over multiple modes, (2) competitive reasoning density within each mode, and (3) integration of these frontier policies into a single unified student model while preserving clear mode separation and high per-mode performance.

cs.LG↗

arXiv2026

CodeSpecBench: Benchmarking LLMs for Executable Behavioral Specification Generation

Large language models (LLMs) can generate code from natural language, but the extent to which they capture intended program behavior remains unclear. Executable behavioral specifications, defined via preconditions and postconditions, provide a concrete means to assess such understanding. However, existing work on specification generation is constrained in evaluation methodology, task settings, and specification expressiveness. We introduce CodeSpecBench, a benchmark for executable behavioral specification generation under an execution-based evaluation protocol. CodeSpecBench supports both function-level and repository-level tasks and encodes specifications as executable Python functions. Constructed from diverse real-world codebases, it enables a realistic assessment of both correctness (accepting valid behaviors) and completeness (rejecting invalid behaviors). Evaluating 15 state-of-the-art LLMs on CodeSpecBench, we observe a sharp performance degradation on repository-level tasks, where the best model attains only a 20.2% pass rate. We further find that specification generation is substantially more challenging than code generation, indicating that strong coding performance does not necessarily reflect deep understanding of intended program semantics. Our data and code are available at https://github.com/SparksofAGI/CodeSpecBench.

cs.SE↗

arXiv2026

A Survey of Mamba

As one of the most representative DL techniques, Transformer architecture has empowered numerous advanced models, especially the large language models (LLMs) that comprise billions of parameters, becoming a cornerstone in deep learning. Despite the impressive achievements, Transformers still face inherent limitations, particularly the time-consuming inference resulting from the quadratic computation complexity of attention calculation. Recently, a novel architecture named Mamba, drawing inspiration from classical state space models (SSMs), has emerged as a promising alternative for building foundation models, delivering comparable modeling abilities to Transformers while preserving near-linear scalability concerning sequence length. This has sparked an increasing number of studies actively exploring Mamba's potential to achieve impressive performance across diverse domains. Given such rapid evolution, there is a critical need for a systematic review that consolidates existing Mamba-empowered models, offering a comprehensive understanding of this emerging model architecture. In this survey, we therefore conduct an in-depth investigation of recent Mamba-associated studies, covering three main aspects: the advancements of Mamba-based models, the techniques of adapting Mamba to diverse data, and the applications where Mamba can excel. Specifically, we first review the foundational knowledge of various representative deep learning models and the details of Mamba-1&2 as preliminaries. Then, to showcase the significance of Mamba for AI, we comprehensively review the related studies focusing on Mamba models' architecture design, data adaptability, and applications. Finally, we present a discussion of current limitations and explore various promising research directions to provide deeper insights for future investigations.

cs.LG↗

arXiv2026

AppellateGen: A Benchmark for Appellate Legal Judgment Generation

Legal judgment generation is a critical task in legal intelligence. However, existing research in legal judgment generation has predominantly focused on first-instance trials, relying on static fact-to-verdict mappings while neglecting the dialectical nature of appellate (second-instance) review. To address this, we introduce AppellateGen, a benchmark for second-instance legal judgment generation comprising 7,351 case pairs. The task requires models to draft legally binding judgments by reasoning over the initial verdict and evidentiary updates, thereby modeling the causal dependency between trial stages. We further propose a judicial Standard Operating Procedure (SOP)-based Legal Multi-Agent System (SLMAS) to simulate judicial workflows, which decomposes the generation process into discrete stages of issue identification, retrieval, and drafting. Experimental results indicate that while SLMAS improves logical consistency, the complexity of appellate reasoning remains a substantial challenge for current LLMs. The dataset and code are publicly available at: https://anonymous.4open.science/r/AppellateGen-5763.

cs.CY↗

arXiv2026

RRAT J1541+4703: A Rotating Radio Transient Exhibiting Normal Pulsar States

Rotating Radio Transients (RRATs) are a class of pulsar-like objects characterized by intermittent radio emissions. Among them, RRATs that exhibit both RRAT and normal pulsar (NP) states may represent a key evolutionary stage from nulling pulsars to RRATs. We performed a detailed analysis of RRAT J1574+4703 using the Five-hundred-meter Aperture Spherical Radio Telescope (FAST) at a frequency of 1250 MHz. Our findings indicate that this RRAT spends approximately 98% of its time in the RRAT state, with the remainder spent in an NP state exhibiting nulling behavior. Additionally, we observed distinct integral pulse profiles and polarization properties between the two states, suggesting that they originate from different emission heights and magnetospheric structures. Furthermore, it was observed that the NP states of this RRAT exhibit mode switching, with ~44% of the time spent in the normal mode and ~39% in the abnormal mode. Notably, abnormal modes are predominantly detected at the onset and termination of the NP states. This discrepancy between the modes indicates potential instability in the magnetospheric processes that govern the NP states.

astro-ph.HE↗

arXiv2026

The annular gap model under a rotating dipole field approximation: simulating gamma-ray light curve

A more realistic description of the magnetosphere is crucial for understanding the radiation emitted by pulsars. In this paper, we revisit the annular gap model by employing a rotating dipole field, which is more realistic than the static dipole field, as an approximation of the magnetic structure of the pulsar magnetosphere. Compared with the static dipole field approximation, the open field-line region, including both the core and annular gaps, is significantly enlarged, and the two regions become asymmetric with respect to the fiducial plane. We apply this model to three young gamma-ray pulsars with distinct light-curve morphologies, PSRs J0631$+$1036 (single peak), J1709$-$4429 (double peaks), and J1048$-$5832 (three peaks). Using viewing geometries constrained by radio polarization measurements, the annular gap model within the rotating dipole field successfully reproduces the main morphological features of their gamma-ray light curves above 0.1 GeV. Our model provides a framework for interpreting pulsar high-energy emission, which can be used to analyze the emission properties of high-energy pulsars.

astro-ph.HE↗

arXiv2026

Asymptotics of the principal eigenvalue of an elliptic operator on closed and orientable Riemannian manifolds

This paper investigates the asymptotic behavior of the principal eigenvalue $λ(s)$, as $s\to+\infty$, for the following elliptic eigenvalue problem \begin{equation*}\label{E} -Δ_{M}u-s\langle \nabla_M f, \nabla_M u\rangle_g +c u=λ(s)u, \end{equation*} defined on an orientable and closed Riemannian manifold $(M,g)$. Assuming $f$ is a Morse function defined on $M$, we find that the limit $\lim\limits_{s\to+\infty} λ(s)$ is determined by the minimum value of the function $c$ over the set of the maximum points of $f$, a result that is independent of the curvature of manifold.

math.AP↗

arXiv2026

TextOVSR: Text-Guided Real-World Opera Video Super-Resolution

Many classic opera videos exhibit poor visual quality due to the limitations of early filming equipment and long-term degradation during storage. Although real-world video super-resolution (RWVSR) has achieved significant advances in recent years, directly applying existing methods to degraded opera videos remains challenging. The difficulties are twofold. First, accurately modeling real-world degradations is complex: simplistic combinations of classical degradation kernels fail to capture the authentic noise distribution, while methods that extract real noise patches from external datasets are prone to style mismatches that introduce visual artifacts. Second, current RWVSR methods, which rely solely on degraded image features, struggle to reconstruct realistic and detailed textures due to a lack of high-level semantic guidance. To address these issues, we propose a Text-guided Dual-Branch Opera Video Super-Resolution (TextOVSR) network, which introduces two types of textual prompts to guide the super-resolution process. Specifically, degradation-descriptive text, derived from the degradation process, is incorporated into the negative branch to constrain the solution space. Simultaneously, content-descriptive text is incorporated into a positive branch and our proposed Text-Enhanced Discriminator (TED) to provide semantic guidance for enhanced texture reconstruction. Furthermore, we design a Degradation-Robust Feature Fusion (DRF) module to facilitate cross-modal feature fusion while suppressing degradation interference. Experiments on our OperaLQ benchmark show that TextOVSR outperforms state-of-the-art methods both qualitatively and quantitatively. The code is available at https://github.com/ChangHua0/TextOVSR.

cs.CV↗

arXiv2026

FedBPrompt: Federated Domain Generalization Person Re-Identification via Body Distribution Aware Visual Prompts

Federated Domain Generalization for Person Re-Identification (FedDG-ReID) learns domain-invariant representations from decentralized data. While Vision Transformer (ViT) is widely adopted, its global attention often fails to distinguish pedestrians from high similarity backgrounds or diverse viewpoints -- a challenge amplified by cross-client distribution shifts in FedDG-ReID. To address this, we propose Federated Body Distribution Aware Visual Prompt (FedBPrompt), introducing learnable visual prompts to guide Transformer attention toward pedestrian-centric regions. FedBPrompt employs a Body Distribution Aware Visual Prompts Mechanism (BAPM) comprising: Holistic Full Body Prompts to suppress cross-client background noise, and Body Part Alignment Prompts to capture fine-grained details robust to pose and viewpoint variations. To mitigate high communication costs, we design a Prompt-based Fine-Tuning Strategy (PFTS) that freezes the ViT backbone and updates only lightweight prompts, significantly reducing communication overhead while maintaining adaptability. Extensive experiments demonstrate that BAPM effectively enhances feature discrimination and cross-domain generalization, while PFTS achieves notable performance gains within only a few aggregation rounds. Moreover, both BAPM and PFTS can be easily integrated into existing ViT-based FedDG-ReID frameworks, making FedBPrompt a flexible and effective solution for federated person re-identification. The code is available at https://github.com/leavlong/FedBPrompt.

cs.CV↗

arXiv2026

FedARKS: Federated Aggregation via Robust and Discriminative Knowledge Selection and Integration for Person Re-identification

The application of federated domain generalization in person re-identification (FedDG-ReID) aims to enhance the model's generalization ability in unseen domains while protecting client data privacy. However, existing mainstream methods typically rely on global feature representations and simple averaging operations for model aggregation, leading to two limitations in domain generalization: (1) Using only global features makes it difficult to capture subtle, domain-invariant local details (such as accessories or textures); (2) Uniform parameter averaging treats all clients as equivalent, ignoring their differences in robust feature extraction capabilities, thereby diluting the contributions of high quality clients. To address these issues, we propose a novel federated learning framework, Federated Aggregation via Robust and Discriminative Knowledge Selection and Integration (FedARKS), comprising two mechanisms: RK (Robust Knowledge) and KS (Knowledge Selection).

cs.CV↗

arXiv2026

Progressive Residual Warmup for Language Model Pretraining

Transformer architectures serve as the backbone for most modern Large Language Models, therefore their pretraining stability and convergence speed are of central concern. Motivated by the logical dependency of sequentially stacked layers, we propose Progressive Residual Warmup (ProRes) for language model pretraining. ProRes implements an "early layer learns first" philosophy by multiplying each layer's residual with a scalar that gradually warms up from 0 to 1, with deeper layers taking longer warmup steps. In this way, deeper layers wait for early layers to settle into a more stable regime before contributing to learning. We demonstrate the effectiveness of ProRes through pretraining experiments across various model scales, as well as normalization and initialization schemes. Comprehensive analysis shows that ProRes not only stabilizes pretraining but also introduces a unique optimization trajectory, leading to faster convergence, stronger generalization and better downstream performance. Our code is available at https://github.com/dandingsky/ProRes.

cs.CL↗

arXiv2026

Perception-R1: Advancing Multimodal Reasoning Capabilities of MLLMs via Visual Perception Reward

Enhancing the multimodal reasoning capabilities of Multimodal Large Language Models (MLLMs) is a challenging task that has attracted increasing attention in the community. Recently, several studies have applied Reinforcement Learning with Verifiable Rewards (RLVR) to the multimodal domain in order to enhance the reasoning abilities of MLLMs. However, these works largely overlook the enhancement of multimodal perception capabilities in MLLMs, which serve as a core prerequisite and foundational component of complex multimodal reasoning. Through McNemar's test, we find that existing RLVR method fails to effectively enhance the multimodal perception capabilities of MLLMs, thereby limiting their further improvement in multimodal reasoning. To address this limitation, we propose Perception-R1, which introduces a novel visual perception reward that explicitly encourages MLLMs to perceive the visual content accurately, thereby can effectively incentivizing both their multimodal perception and reasoning capabilities. Specifically, we first collect textual visual annotations from the CoT trajectories of multimodal problems, which will serve as visual references for reward assignment. During RLVR training, we employ a judging LLM to assess the consistency between the visual annotations and the responses generated by MLLM, and assign the visual perception reward based on these consistency judgments. Extensive experiments on several multimodal reasoning benchmarks demonstrate the effectiveness of our Perception-R1, which achieves state-of-the-art performance on most benchmarks using only 1,442 training data. Our code and dataset will be available at https://github.com/tongxiao2002/Perception-R1.

cs.LG↗

arXiv2026

CSyMR: Benchmarking Compositional Music Information Retrieval in Symbolic Music Reasoning

Natural language information needs over symbolic music scores rarely reduce to a single step lookup. Many queries require compositional Music Information Retrieval (MIR) that extracts multiple pieces of evidence from structured notation and aggregates them to answer the question. This setting remains challenging for Large Language Models due to the mismatch between natural language intents and symbolic representations, as well as the difficulty of reliably handling long structured contexts. Existing benchmarks only partially capture these retrieval demands, often emphasizing isolated theoretical knowledge or simplified settings. We introduce CSyMR-Bench, a benchmark for compositional MIR in symbolic music reasoning grounded in authentic user scenarios. It contains 126 multiple choice questions curated from community discussions and professional examinations, where each item requires chaining multiple atomic analyses over a score to derive implicit musical evidence. To support diagnosis, we provide a taxonomy with six query intent categories and six analytical dimension tags. We further propose a tool-augmented retrieval and reasoning framework that integrates a ReAct-style controller with deterministic symbolic analysis operators built with music21. Experiments across prompting baselines and agent variants show that tool-grounded compositional retrieval consistently outperforms Large Language Model-only approaches, yielding 5-7% absolute accuracy gains, with the largest improvements on analysis-heavy categories.

cs.LG↗

arXiv2026

From Calibration to Refinement: Seeking Certainty via Probabilistic Evidence Propagation for Noisy-Label Person Re-Identification

With the increasing demand for robust person Re-ID in unconstrained environments, learning from datasets with noisy labels and sparse per-identity samples remains a critical challenge. Existing noise-robust person Re-ID methods primarily rely on loss-correction or sample-selection strategies using softmax outputs. However, these methods suffer from two key limitations: 1) Softmax exhibits translation invariance, leading to over-confident and unreliable predictions on corrupted labels. 2) Conventional sample selection based on small-loss criteria often discards valuable hard positives that are crucial for learning discriminative features. To overcome these issues, we propose the CAlibration-to-REfinement (CARE) method, a two-stage framework that seeks certainty through probabilistic evidence propagation from calibration to refinement. In the calibration stage, we propose the probabilistic evidence calibration (PEC) that dismantles softmax translation invariance by injecting adaptive learnable parameters into the similarity function, and employs an evidential calibration loss to mitigate overconfidence on mislabeled samples. In the refinement stage, we design the evidence propagation refinement (EPR) that can more accurately distinguish between clean and noisy samples. Specifically, the EPR contains two steps: Firstly, the composite angular margin (CAM) metric is proposed to precisely distinguish clean but hard-to-learn positive samples from mislabeled ones in a hyperspherical space; Secondly, the certainty-oriented sphere weighting (COSW) is developed to dynamically allocate the importance of samples according to CAM, ensuring clean instances drive model updates. Extensive experimental results on Market1501, DukeMTMC-ReID, and CUHK03 datasets under both random and patterned noises show that CARE achieves competitive performance.

cs.CV↗

arXiv2026

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↗

arXiv2026

Learning Positive-Incentive Point Sampling in Neural Implicit Fields for Object Pose Estimation

Learning neural implicit fields of 3D shapes is a rapidly emerging field that enables shape representation at arbitrary resolutions. Due to the flexibility, neural implicit fields have succeeded in many research areas, including shape reconstruction, novel view image synthesis, and more recently, object pose estimation. Neural implicit fields enable learning dense correspondences between the camera space and the object's canonical space-including unobserved regions in camera space-significantly boosting object pose estimation performance in challenging scenarios like highly occluded objects and novel shapes. Despite progress, predicting canonical coordinates for unobserved camera-space regions remains challenging due to the lack of direct observational signals. This necessitates heavy reliance on the model's generalization ability, resulting in high uncertainty. Consequently, densely sampling points across the entire camera space may yield inaccurate estimations that hinder the learning process and compromise performance. To alleviate this problem, we propose a method combining an SO(3)-equivariant convolutional implicit network and a positive-incentive point sampling (PIPS) strategy. The SO(3)-equivariant convolutional implicit network estimates point-level attributes with SO(3)-equivariance at arbitrary query locations, demonstrating superior performance compared to most existing baselines. The PIPS strategy dynamically determines sampling locations based on the input, thereby boosting the network's accuracy and training efficiency. Our method outperforms the state-of-the-art on three pose estimation datasets. Notably, it demonstrates significant improvements in challenging scenarios, such as objects captured with unseen pose, high occlusion, novel geometry, and severe noise.

cs.CV↗

arXiv2026

On Predictability of Reinforcement Learning Dynamics for Large Language Models

Recent advances in reasoning capabilities of large language models (LLMs) are largely driven by reinforcement learning (RL), yet the underlying parameter dynamics during RL training remain poorly understood. This work identifies two fundamental properties of RL-induced parameter updates in LLMs: (1) Rank-1 Dominance, where the top singular subspace of the parameter update matrix nearly fully determines reasoning improvements, recovering over 99\% of performance gains; and (2) Rank-1 Linear Dynamics, where this dominant subspace evolves linearly throughout training, enabling accurate prediction from early checkpoints. Extensive experiments across 8 LLMs and 7 algorithms validate the generalizability of these properties. More importantly, based on these findings, we propose AlphaRL, a plug-in acceleration framework that extrapolates the final parameter update using a short early training window, achieving up to 2.5 speedup while retaining \textgreater 96\% of reasoning performance without extra modules or hyperparameter tuning. This positions our finding as a versatile and practical tool for large-scale RL, opening a path toward principled, interpretable, and efficient training paradigm for LLMs.

cs.LG↗

arXiv2026

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↗