Searcharxiv⌕ Search

arXiv subjects

Zhe Xu

Publications and source records attributed to Zhe Xu.

At least 37 records · Page 2Linked to original sources
arXiv2026

Self-Guided Test-Time Training for Long-Context LLMs

Long-context processing has become increasingly important for large language models (LLMs), but simply extending the context window does not guarantee effective utilization of long inputs. As input length grows, accuracy often degrades, indicating that models still struggle to identify and use the evidence most relevant to a question. A promising way to improve long-context utilization is test-time training (TTT), which treats the test context as a training example for instance-specific parameter adaptation. However, applying TTT to the entire long context is prohibitively expensive, while adapting on randomly sampled spans introduces severe noise. Because most spans in a long context are irrelevant to the specific question, training on them may even degrade the base model's performance. Our preliminary study shows that TTT is highly sensitive to training-span quality: on LongBench-v2, TTT on randomly sampled spans hurts performance, whereas TTT on oracle spans substantially improves it. Motivated by this, we propose a simple method, Self-Guided TTT (S-TTT): before adaptation, the model identifies the evidence spans it should learn from, and the standard language-modeling training objective is applied only to those selected spans. On two challenging long-context reasoning benchmarks, LongBench-v2 and LongBench-Pro, S-TTT improves accuracy for both Qwen3-4B-Thinking-2507 and Llama-3.1-8B-Instruct, achieving up to a 15% relative improvement.

cs.CL↗
arXiv2026

Observational Evidence of Solar Spicules Associated with Microfilament Eruptions Using DKIST

The formation mechanism of spicules is fundamentally important for understanding mass and energy transport from the chromosphere into the corona. Recent studies suggested that spicules may be powered by microfilament eruptions. However, direct observational evidence remains limited due to insufficient spatial resolution. Using high-resolution H$α$ broadband observations from the Visible Broadband Imager (VBI) onboard the Daniel K. Inouye Solar Telescope (DKIST), we identify 30 spicule events triggered by microfilament eruptions in a quiet Sun region near the solar disk center on 2023 August 29. The detected microfilaments have an average length of $0.93\pm0.46$ Mm and a minimum length of 0.17 Mm, substantially smaller than previously reported minifilaments. We identify two distinct morphological classes of ejecta: individual spicules associated with smaller microfilaments, and enhanced spicular activities associated with larger microfilaments. Moreover, some events exhibit apparent twisting motions. All these high-resolution observations provide compelling evidence that spicules can be triggered by microfilament eruptions.

astro-ph.SR↗
arXiv2026

Analytical solution and Lie algebra of the relativistic Boltzmann equation

In this work, we present a novel and more efficient approach to constructing the relativistic Bobylev, Krook, and Wu (BKW) solution [A. V. Bobylev, Sov. Phys. Dokl. 20, 820 (1976); M. Krook and T. T. Wu, Phys. Fluids 20, 1589 (1977)] of the relativistic Boltzmann equation. Introducing a new ansatz for the distribution function, we demonstrate that within this specific ansatz space, only the equilibrium and BKW-type forms yield exact solutions to the nonlinear Boltzmann equation. Furthermore, we also derive the Lie algebra of invariant transformations admitted by the relativistic Boltzmann equation and show that the corresponding symmetry group transformations can be systematically constructed from this algebra.

nucl-th↗
arXiv2026

Democratizing and accelerating AI-driven pathology research through agentic intelligence

Computational pathology has advanced rapidly with the emergence of foundation models, yet widespread adoption remains limited by substantial technical complexity and programming requirements. Here we present PathLab, an autonomous agentic framework that translates natural-language research objectives into executable and validated computational pathology workflows through the structured composition of domain-specific skills and tools. By organizing workflow generation around reusable methodological modules, including data preprocessing, model development, evaluation and interpretation, PathLab enables studies to be specified at the level of scientific intent rather than implementation details. We evaluated PathLab across 12 public datasets spanning four representative task families: region-of-interest classification, whole-slide image classification, segmentation and survival prediction. Across all task categories, PathLab achieved non-inferior performance relative to expert implementations, while consistently enforcing semantic validity of user prompts and proactively rejecting incompatible workflow specifications prior to execution. In controlled user studies, PathLab substantially reduced the time required to generate executable analytical pipelines and enabled domain experts without programming experience to independently design, execute and evaluate computational pathology studies. Together, these results establish PathLab as a reliable interface between biomedical intent and computational execution, enabling computational pathology studies to be designed at the level of scientific questions rather than programming expertise. By lowering technical barriers to advanced AI methodologies, PathLab provides a foundation for the broader democratization of computational pathology.

cs.AI↗
arXiv2026

Blind Dexterous Grasping via Real2Sim2Real Tactile Policy Learning

Blind grasping with a dexterous hand is a crucial manipulation capability. Nevertheless, learning such tactile-only policies for real robots remains challenging due to the tactile sim-to-real gap and the limited expressiveness of sparse tactile signals. To bridge this gap, we propose a framework for tactile-only blind grasping that is deployable on a physical multi-fingered robotic hand. Our approach combines three key components. First, we introduce a Real2Sim tactile calibration pipeline that constructs a contact-calibrated digital-twin simulator capable of reproducing real tactile signals. Second, we improve the expressiveness of sparse tactile observations using a layout-aware tactile encoder, which incorporates sensor-geometry priors through self-supervised pretraining. Third, to improve generalization to unseen objects, we train object-specific reinforcement-learning experts in the calibrated simulator and aggregate their successful grasp trajectories into a tactile-conditioned Diffusion Policy. We evaluate our method on a physical LEAP Hand equipped with distributed tactile sensing across 10 seen and 10 unseen objects. The deployed policy achieves a 27\% real-world grasp success rate across all 20 objects, without real-world grasping demonstrations or visual input. Simulation ablations show that layout-aware tactile pretraining improves grasping performance, while sensing-level evaluations confirm that Real2Sim calibration increases the consistency of tactile contact events between simulation and hardware. Together, these results suggest that contact-event calibration, geometry-aware tactile representation learning, and diffusion-based policy aggregation provide an effective path toward tactile-only blind grasping on real dexterous robotic hands. Project page:Dex-Blind-Grasp.github.io.

cs.RO↗
arXiv2026

Two Bridges, One Pathway: From VLMs to Generalizable VLAs with Embodied Trajectory-Coupled Data

Vision-language models (VLMs) are powerful general-purpose reasoners, yet converting them into robot control policies (VLAs) is surprisingly difficult. The root cause is a two-fold gap: VLMs are trained on internet-scale images with language-understanding objectives, while VLAs must perceive robot scenes and predict motor actions. Fine-tuning a VLM directly on robot action data forces the model to cross both gaps at once -- the learning curve is steep and the rich generalizations learned during pretraining tend to degrade rather than transfer. We argue that this gap can be bridged gradually with the right intermediate data. We introduce \emph{embodied trajectory-coupled (ETC) data} -- vision-language supervision derived from the same robot scenes and trajectories used for action learning. Because ETC data shares the visual context of robot operation while retaining familiar language-understanding objectives, it provides a natural stepping stone between VLM pretraining and VLA fine-tuning. Building on this, we design a three-stage training recipe. Distribution Bridging first adapts the VLM to embodied visual-language semantics. Objective Bridging then gradually shifts the model toward action prediction while preserving the acquired representations. Retentive Adaptation finally specializes the policy to the target deployment domain. We further show that mixing task-relevant out-of-distribution ETC data with a small amount of action data enables the model to generalize to novel visual-language conditions without requiring additional robot demonstrations. Simulation and real-robot experiments confirm that this gradual bridging strategy is the key to transferring VLM generalization into robust, deployable robot policies.

cs.RO↗
arXiv2026

MOSS-Audio Technical Report

MOSS-Audio is a unified audio-language model for speech, environmental sound, and music understanding, supporting audio captioning, time-aware question answering, timestamped transcription, and audio-grounded reasoning. MOSS-Audio couples a dedicated audio encoder with a modality adapter and a large language model: the encoder produces 12.5 Hz temporal representations, the adapter projects them into the decoder space, and the decoder generates autoregressive text outputs. Two design choices are central to the system: DeepStack cross-layer feature injection, which exposes the decoder to acoustic information from multiple encoder depths, and time markers, which provide explicit temporal cues by inserting timestamp markers into the audio-token stream. At the data level, we design an event-preserving audio annotation pipeline that segments raw audio at coherent event boundaries, applies branch-specific annotation to speech, music, and general audio, and merges the results into unified captions for pretraining. The intermediate branch-specific captions are further retained to support the construction of task-oriented SFT data. The model is pretrained on large-scale audio-language data, with time-aware objectives incorporated to support temporal grounding, and then undergoes multi-stage post-training to enhance instruction following and audio-grounded reasoning. We release 4B and 8B variants in both Instruct and Thinking configurations. MOSS-Audio achieves strong performance across general audio understanding, speech captioning, ASR, and timestamped ASR, positioning it as a promising understanding foundation for future voice agents.

cs.SD↗
arXiv2026

Beyond Literal Translation: Evaluating Cultural Effectiveness in Social Media UGC

Social media platforms enable large-scale cross-lingual communication, but translating user-generated content (UGC) remains challenging due to its informal style, cultural references, and interaction-based expressions. While recent LLMs have improved translation quality, existing benchmarks and metrics often fail to capture whether translations convey intended meaning and cultural resonance in real-world settings. In this work, we introduce CULTURE-MT, a benchmark for social media translation that focuses on both CULtural Transmission and UGC-specific emotion REsonance. CULTURE-MT consists of 1,002 UGC notes across 14 domains, categorized into four types based on culture-loaded symbols and linguistic style features. We also construct UGC-oriented training data to fine-tune Qwen3-8B and Qwen3-32B as baselines. We propose cultural effectiveness as a new evaluation criterion, focusing on expression accuracy and cultural adaptability. Testing 15 models, including the baselines, we find that traditional metrics fail to capture cultural effectiveness. We also observe that cultural effectiveness on base LLMs correlates with model size. Our work provides a comprehensive evaluation system for UGC translation models and will offer an open evaluation platform to advance research in this area. We release the CULTURE-MT benchmark and provide an online leaderboard where submitted translation results can be evaluated by our trained JUDGER.

cs.CL↗
arXiv2026

HCInfer: An Efficient Inference System via Error Compensation for Resource-Constrained Devices

LLMs often struggle with memory-constrained deployment on consumer-grade hardware due to their massive parameter sizes. While existing solutions such as model compression and offloading improve deployment feasibility, they often suffer from substantial accuracy degradation or severe throughput bottlenecks. Recent error compensation methods recover accuracy through auxiliary LoRA-style branches, and we observe that these branches are inherently amenable to offloading: they require substantial parameter storage but access only a small subset of compensation parameters during each inference step. Motivated by this opportunity, we propose HCInfer, a heterogeneous inference system that offloads residual compensation to the CPU while executing the compressed backbone on the GPU, and further introduces an asynchronous compensation pipeline and sensitivity-aware dynamic rank allocation to hide compensation overhead and maximize accuracy recovery. Experimental results show that HCInfer achieves a maximum accuracy improvement of 5.2% on downstream tasks compared to compression model and sustaining a maximum speedup of 10.4x compared to full-precision model.

cs.LG↗
arXiv2026

Perceive, Verify and Understand Long Video: Multi-Granular Perception and Active Verification via Interactive Agents

Long videos, characterized by temporal complexity and sparse task-relevant information, pose significant reasoning challenges for AI systems. Although existing Large Language Model (LLM)-based approaches have advanced long video understanding, they remain bottlenecked by task-agnostic, fixed-granularity perception pipelines and suffer from vision-language hallucinations. Inspired by human adaptive perception and active verification, we propose CogniGPT, a framework leveraging an interactive loop between a Multi-Granular Perception Agent (MPA) and an Active Verification Agent (AVA). Specifically, instead of predetermined heuristics, MPA adaptively determines the optimal perception granularity and strategy based on the evolving context, while AVA actively mines multi-perspective visual evidence to cross-verify key observations and eliminate hallucinations. This interaction allows CogniGPT to efficiently identify a minimal set of reliable task-related clues. Extensive experiments on EgoSchema, Video-MME, NExT-QA, and MovieChat demonstrate its superiority in accuracy and efficiency. Notably, on EgoSchema, it surpasses existing training-free methods using only 11.2 frames and achieves performance comparable to Gemini 1.5-Pro.

cs.CV↗
arXiv2026

Fast Strategy Solving for the Informed Player in Two-Player Zero-Sum Linear-Quadratic Differential Games with One-Sided Information

We study finite-horizon two-player zero-sum differential games with one-sided payoff information ($G$), where the informed player (P1) knows the game payoff, while P2 only has a public belief over a finite set of possible payoffs. In this case, P1's Nash equilibrium (NE) behavioral strategy may control the release of the type information or even resort to manipulate P2's belief. Previous studies revealed an atomic structure of the NE of $G$ with general nonlinear dynamics and payoffs, leading to tractable NE approximation. Implementing such approximation schemes for real-time sub-game solving, however, has not been achieved, yet is desired for applications where sim-to-real gaps exist and robust control is required. This paper improves the computational efficiency of sub-game solving for P1 during $G$ with linear dynamics and quadratic losses. Specifically, we show that P1's NE computation can be formulated as a bi-level optimization problem where the outer level optimizes the "signaling" strategy, i.e., when and how to reveal information through control, and the inner level is a game-tree LQR that solves for the optimal closed-loop control. This bi-level problem is solved via an adjoint-enabled backpropagation scheme: A "backward" LQR pass is followed by a "forward" gradient descent pass for improving the signaling. We apply the proposed algorithm to approximate NEs for variants of a homing problem with a 8D state space, 2D action spaces, and a discrete time horizon of $K=10$. The algorithm achieves $\approx$10Hz sub-game solving, enabling robust game-theoretic planning under information asymmetry and random disturbances.

cs.GT↗
arXiv2026

Repeated Sunspot Light Bridge Jets Associated with Slipping Base Brightenings

Light bridge (LB) jets offer a unique window into small-scale eruptive phenomena within sunspots, the Sun's strongest magnetic environments; however, their generation mechanism remains a subject of debate. Using high-resolution observations from the New Vacuum Solar Telescope (NVST), we investigated six recurrent light bridge jets and the slipping motions of their jet base points (JBPs). Analogous to coronal jets, our observations show that these LB jets are characterized by a preceding JBP followed by a collimated jet spire. The JBP of each repeated jet along the LB displays apparent slipping motion at velocities of 0.6-1.5 km/s, which is temporally correlated with quasi-periodic enhanced photospheric horizontal motion of 1.3-6.5 km/s. Following the slipping JBPs, the resulting jet spires' fronts display similar slipping behaviors within the upper solar atmosphere. The Chinese Ha Solar Explorer (CHASE) reveals Ellerman-bomb-like spectral signatures at the JBPs, confirming that magnetic reconnection is operating at the jet base. Based on these results, we propose that repeated 3D reconnection occurring between the horizontal LB field and the ambient vertical umbral field may drive these LB jets. This process appears to be driven and/or modulated by quasi-periodic horizontal motion fueled by convective upflows and the transport of magnetic flux along the light bridge. This work suggests that some LB jets share a common reconnection-driven mechanism with coronal jets and provides direct evidence of slipping reconnection occurring along the sunspot light bridge.

astro-ph.SR↗
arXiv2026

ETac: A Lightweight and Efficient Tactile Simulation Framework for Learning Dexterous Manipulation

Tactile sensors are increasingly integrated into dexterous robotic manipulators to enhance contact perception. However, learning manipulation policies that rely on tactile sensing remains challenging, primarily due to the trade-off between fidelity and computational cost of soft-body simulations. To address this, we present ETac, a tactile simulation framework that models elastomeric soft-body interactions with both high fidelity and efficiency. ETac employs a lightweight data-driven deformation propagation model to capture soft-body contact dynamics, achieving high simulation quality and boosting efficiency that enables large-scale policy training. When serving as the simulation backend, ETac produces surface deformation estimates comparable to FEM and demonstrates applicability for modeling real tactile sensors. Then, we showcase its capability in training a blind grasping policy that leverages large-area tactile feedback to manipulate diverse objects. Running on a single RTX 4090 GPU, ETac supports reinforcement learning across 4,096 parallel environments, achieving a total throughput of 869 FPS. The resulting policy reaches an average success rate of 84.45% across four object types, underscoring ETac's potential to make tactile-based skill learning both efficient and scalable.

cs.RO↗
arXiv2026

NTIRE 2026 The 3rd Restore Any Image Model (RAIM) Challenge: AI Flash Portrait (Track 3)

In this paper, we present a comprehensive overview of the NTIRE 2026 3rd Restore Any Image Model (RAIM) challenge, with a specific focus on Track 3: AI Flash Portrait. Despite significant advancements in deep learning for image restoration, existing models still encounter substantial challenges in real-world low-light portrait scenarios. Specifically, they struggle to achieve an optimal balance among noise suppression, detail preservation, and faithful illumination and color reproduction. To bridge this gap, this challenge aims to establish a novel benchmark for real-world low-light portrait restoration. We comprehensively evaluate the proposed algorithms utilizing a hybrid evaluation system that integrates objective quantitative metrics with rigorous subjective assessment protocols. For this competition, we provide a dataset containing 800 groups of real-captured low-light portrait data. Each group consists of a 1K-resolution low-light input image, a 1K ground truth (GT), and a 1K person mask. This challenge has garnered widespread attention from both academia and industry, attracting over 100 participating teams and receiving more than 3,000 valid submissions. This report details the motivation behind the challenge, the dataset construction process, the evaluation metrics, and the various phases of the competition. The released dataset and baseline code for this track are publicly available from the same \href{https://github.com/zsn1434/AI_Flash-BaseLine/tree/main}{GitHub repository}, and the official challenge webpage is hosted on \href{https://www.codabench.org/competitions/12885/}{CodaBench}.

cs.CV↗
arXiv2026

Spectroscopic Case Studies of Four Long-duration Transition-region Explosive Events

This work presents a detailed spectroscopic case study of four long-duration transition-region (TR) explosive events (EEs) observed in NOAA Active Region 13213 on 2023 February 10 using the Interface Region Imaging Spectrograph. The dynamic spectral evolution of each event is tracked through multicomponent Gaussian fitting of the Si IV 1403 Å line profiles. Three recurrent spectral morphologies are identified and characterized: bilateral wing enhancement, exclusive red-wing enhancement, and exclusive blue-wing enhancement, among which bilateral enhancement is the most common in the studied cases. Throughout their lifetimes of 20-25 minutes, these events display sustained and evolving bidirectional flows, with high-velocity components ($|v|$ > 100 km $s^{-1}$) emerging in late phases. These spectral signatures are interpreted as evidence of ongoing or recurrent magnetic reconnection, where bilateral profiles correspond to bidirectional outflows, and exclusive wing enhancements represent geometric or evolutionary phases of the same process. In contrast, cotemporal flare ribbons and loop structures exhibit pronounced, unidirectional redshifts. This study underscores that significant non-Gaussian wing enhancement, rather than exclusively high speed, constitutes a defining spectroscopic signature of EEs, and provides detailed kinematic constraints on the dynamics of such TR EEs.

astro-ph.SR↗
arXiv2026

Bi-CRCL: Bidirectional Conservative-Radical Complementary Learning with Pre-trained Foundation Models for Class-incremental Medical Image Analysis

Class-incremental learning (CIL) in medical image-guided diagnosis requires retaining prior diagnostic knowledge while adapting to newly emerging disease categories, which is critical for scalable clinical deployment. This problem is particularly challenging due to heterogeneous data and privacy constraints that prevent memory replay. Although pretrained foundation models (PFMs) have advanced general-domain CIL, their potential in medical imaging remains underexplored, where domain-specific adaptation is essential yet difficult due to anatomical complexity and inter-institutional heterogeneity. To address this gap, we conduct a systematic benchmark of recent PFM-based CIL methods and propose Bidirectional Conservative-Radical Complementary Learning (Bi-CRCL), a dual-learner framework inspired by complementary learning systems. Bi-CRCL integrates a conservative learner that preserves prior knowledge through stability-oriented updates and a radical learner that rapidly adapts to new categories via plasticity-oriented learning. A bidirectional interaction mechanism enables forward transfer and backward consolidation, allowing continual integration of new knowledge while mitigating catastrophic forgetting. During inference, outputs from both learners are adaptively fused for robust predictions. Experiments on five medical imaging datasets demonstrate consistent improvements over state-of-the-art methods under diverse settings, including cross-dataset shifts and varying task configurations.

cs.CV↗
arXiv2026

Quasiperiodic Slipping Motion of Flare Ribbon Fine Structures Anchored in a Sunspot Light Bridge

We used high-resolution observations from the New Vacuum Solar Telescope and the Solar Dynamics Observatory to carry out a detailed multiwavelength analysis of the fine structures in the flare ribbon of a C3.9-class flare on 22 April 2021. A segment of the flare ribbon was rooted in a sunspot light bridge and exhibited discrete substructures, which we term "burrs", with equivalent diameters of 233-895 km and inter-core separations of 1129-1739 km. These structures are characterized by discrete redshifted cores accompanied by "tails" with lengths of 700-1370 km and widths of 310-600 km that show faint blueshifts. The burrs display systematic slipping motions along the ribbon, with apparent velocities decreasing from about 40 to 21 km/s, and show a distinct quasi-periodicity of about 6 minutes in H-alpha and EUV passbands. Differential emission measure analysis indicates that the emitting plasma is multi-thermal and dominated by temperatures of 1-2 MK. The observed morphology and kinematics are consistent with impulsive energy deposition by precipitating plasmoids, or oblique flux ropes, produced by tearing-mode fragmentation in the coronal current sheet. The close spatiotemporal association between the tails and blueshifts supports the interpretation that these features are related to untwisting magnetic flux ropes. The approximately 6-minute periodicity further suggests that the reconnection process may be modulated by photospheric p-mode oscillations coupled with tearing-mode instability. These results provide observational evidence that light-bridge-anchored fine structures can act as elementary units of flare energy release.

astro-ph.SR↗
arXiv2026

Flow Matching Meets Biology and Life Science: A Survey

Over the past decade, advances in generative modeling, such as generative adversarial networks, masked autoencoders, and diffusion models, have significantly transformed biological research and discovery, enabling breakthroughs in molecule design, protein generation, catalysis discovery, drug discovery, and beyond. At the same time, biological applications have served as valuable testbeds for evaluating the capabilities of generative models. Recently, flow matching has emerged as a powerful and efficient alternative to diffusion-based generative modeling, with growing interest in its application to problems in biology and life sciences. This paper presents the first comprehensive survey of recent developments in flow matching and its applications in biological domains. We begin by systematically reviewing the foundations and variants of flow matching, and then categorize its applications into three major areas: biological sequence modeling, molecule generation and design, and peptide and protein generation. For each, we provide an in-depth review of recent progress. We also summarize commonly used datasets and software tools, and conclude with a discussion of potential future directions. The corresponding curated resources are available at https://github.com/Violet24K/Awesome-Flow-Matching-Meets-Biology.

cs.LG↗