arXiv subjects

Yibo Wang

Publications and source records attributed to Yibo Wang.

At least 109 records · Page 6Linked to original sources

arXiv2025

Ovis2.5 Technical Report

We present Ovis2.5, a successor to Ovis2 designed for native-resolution visual perception and strong multimodal reasoning. Ovis2.5 integrates a native-resolution vision transformer that processes images at their native, variable resolutions, avoiding the degradation from fixed-resolution tiling and preserving both fine detail and global layout -- crucial for visually dense content like complex charts. To strengthen reasoning, we train the model to move beyond linear chain-of-thought and perform reflection -- including self-checking and revision. This advanced capability is exposed as an optional "thinking mode" at inference time, allowing users to trade latency for enhanced accuracy on difficult inputs. The model is trained via a comprehensive five-phase curriculum that progressively builds its skills. The process begins with foundational visual and multimodal pretraining, advances through large-scale instruction tuning, and culminates in alignment and reasoning enhancement using DPO and GRPO. To scale these upgrades efficiently, we employ multimodal data packing and hybrid parallelism, yielding a significant end-to-end speedup. We release two open-source models: Ovis2.5-9B and Ovis2.5-2B. The latter continues the "small model, big performance" philosophy of Ovis2, making it ideal for resource-constrained, on-device scenarios. On the OpenCompass multimodal leaderboard, Ovis2.5-9B averages 78.3, marking a substantial improvement over its predecessor, Ovis2-8B, and achieving state-of-the-art results among open-source MLLMs in the sub-40B parameter range; Ovis2.5-2B scores 73.9, establishing SOTA for its size. Beyond aggregate scores, Ovis2.5 achieves leading results on STEM benchmarks, exhibits strong capabilities on grounding and video tasks, and achieves open-source SOTA at its scale for complex chart analysis.

cs.CV↗

arXiv2025

Metabook: A Mobile-to-Headset Pipeline for 3D Story Book Creation in Augmented Reality

The AR 3D book has shown significant potential in enhancing students' learning outcomes. However, the creation process of 3D books requires a significant investment of time, effort, and specialized skills. Thus, in this paper, we first conduct a three-day workshop investigating how AI can support the automated creation of 3D books. Informed by the design insights derived from the workshop, we developed Metabook, a system that enables even novice users to create 3D books from text automatically. To our knowledge, Metabook is the first system to offer end-to-end 3D book generation. A follow-up study with adult users indicates that Metabook enables inexperienced users to create 3D books, achieving reduced efforts and shortened preparation time. We subsequently recruited 22 children to examine the effects of AR 3D books on children's learning compared with paper-based books. The findings indicate that 3D books significantly enhance children's interest, improve memory retention, and reduce cognitive load, though no significant improvement was observed in comprehension. We conclude by discussing strategies for more effectively leveraging 3D books to support children's learning and offer practical recommendations for educators.

cs.HC↗

arXiv2025

Compact and robust design of the optical system for cold atom interferometer in space

The optical system is a complex and precise subsystem for the atom interferometer (AI), especially for those used in field or space applications. Here, we introduce the design of the optical system of the China Space Station atom interferometer (CSSAI). The scheme is optimized to reduce the complexity while maintaining the capability to achieve the dual-species AI. It features a fused silica optical bench with bonding technology, ensuring compactness and smaller thermal deformation. Spatial structures are designed to isolate the vibration and transfer the heat. After assembling, the optical system has a size of 250 mm * 240 mm * 104 mm and weighs 5.2 kg. After installing in the CSSAI, it passed the thermal and mechanical tests and then launched to the China Space Station (CSS). The output laser power changes are less than 15% from ground to space, and its long-term fluctuations are less than 2.5% for months in space. Cold atom preparation and interference are also realized in space. This optical system is extremely integrated and robust, which provides a foundation for the design of future cold atom payloads in space.

physics.atom-ph↗

arXiv2025

Unveiling the Cosmic Dance of Repeated Nuclear Transient ASASSN-14ko: Insights from Multiwavelength Observations

ASASSN-14ko is a periodically repeating nuclear transient. We conducted high-cadence, multiwavelength observations of this source, revealing several recurrent early bumps and rebrightenings in its UV/optical light curves. The energy released during these bumps and rebrightenings shows a diminishing trend in recent UV/optical outbursts, which we monitored through multiwavelength observations. These features can be ascribed to the interaction between stream debris and the expanded disk in the repeated partial tidal disruption event. The X-ray light curve exhibits an inverse pattern compared to the UV/optical bands, displaying sporadic outbursts. Furthermore, our observations demonstrate that the blackbody temperature and radius in each outburst increase with the UV/optical luminosity, and such evolution resembles that observed in X-ray quasiperiodic eruptions, whereas distinguishing it from typical tidal disruption events.

astro-ph.HE↗

arXiv2025

MCTS-Refined CoT: High-Quality Fine-Tuning Data for LLM-Based Repository Issue Resolution

LLMs demonstrate strong performance in auto-mated software engineering, particularly for code generation and issue resolution. While proprietary models like GPT-4o achieve high benchmarks scores on SWE-bench, their API dependence, cost, and privacy concerns limit adoption. Open-source alternatives offer transparency but underperform in complex tasks, especially sub-100B parameter models. Although quality Chain-of-Thought (CoT) data can enhance reasoning, current methods face two critical flaws: (1) weak rejection sampling reduces data quality, and (2) inadequate step validation causes error accumulation. These limitations lead to flawed reasoning chains that impair LLMs'ability to learn reliable issue resolution. The paper proposes MCTS-REFINE, an enhanced Monte Carlo Tree Search (MCTS)-based algorithm that dynamically validates and optimizes intermediate reasoning steps through a rigorous rejection sampling strategy, generating high-quality CoT data to improve LLM performance in issue resolution tasks. Key innovations include: (1) augmenting MCTS with a reflection mechanism that corrects errors via rejection sampling and refinement, (2) decomposing issue resolution into three subtasks-File Localization, Fault Localization, and Patch Generation-each with clear ground-truth criteria, and (3) enforcing a strict sampling protocol where intermediate outputs must exactly match verified developer patches, ensuring correctness across reasoning paths. Experiments on SWE-bench Lite and SWE-bench Verified demonstrate that LLMs fine-tuned with our CoT dataset achieve substantial improvements over baselines.Notably, Qwen2.5-72B- Instruct achieves 28.3%(Lite) and 35.0%(Verified) resolution rates, surpassing SOTA baseline SWE-Fixer-Qwen-72B with the same parameter scale, which only reached 24.7%(Lite) and 32.8%(Verified).

cs.SE↗

arXiv2025

TestNUC: Enhancing Test-Time Computing Approaches and Scaling through Neighboring Unlabeled Data Consistency

Test-time computing approaches, which leverage additional computational resources during inference, have been proven effective in enhancing large language model performance. This work introduces a novel, linearly scaling approach, TestNUC, that improves test-time predictions by leveraging the local consistency of neighboring unlabeled data-it classifies an input instance by considering not only the model's prediction on that instance but also on neighboring unlabeled instances. We evaluate TestNUC across eight diverse datasets, spanning intent classification, topic mining, domain discovery, and emotion detection, demonstrating its consistent superiority over baseline methods such as standard prompting and self-consistency. Furthermore, TestNUC can be seamlessly integrated with existing test-time computing approaches, substantially boosting their performance. Our analysis reveals that TestNUC scales effectively with increasing amounts of unlabeled data and performs robustly across different embedding models, making it practical for real-world applications. Our code is available at https://github.com/HenryPengZou/TestNUC.

cs.CL↗

arXiv2025

Beyond 'Aha!': Toward Systematic Meta-Abilities Alignment in Large Reasoning Models

Large reasoning models (LRMs) already possess a latent capacity for long chain-of-thought reasoning. Prior work has shown that outcome-based reinforcement learning (RL) can incidentally elicit advanced reasoning behaviors such as self-correction, backtracking, and verification phenomena often referred to as the model's "aha moment". However, the timing and consistency of these emergent behaviors remain unpredictable and uncontrollable, limiting the scalability and reliability of LRMs' reasoning capabilities. To address these limitations, we move beyond reliance on prompts and coincidental "aha moments". Instead, we explicitly align models with three meta-abilities: deduction, induction, and abduction, using automatically generated, self-verifiable tasks. Our three stage-pipeline individual alignment, parameter-space merging, and domain-specific reinforcement learning, boosting performance by over 10\% relative to instruction-tuned baselines. Furthermore, domain-specific RL from the aligned checkpoint yields an additional gain in performance ceiling for both 7B and 32B models across math, coding, and science benchmarks, demonstrating that explicit meta-ability alignment offers a scalable and dependable foundation for reasoning. Code is available at: https://github.com/zhiyuanhubj/Meta-Ability-Alignment

cs.CL↗

arXiv2025

Towards Automated Discovery of Asymmetric Mempool DoS in Blockchains

In blockchains, mempool controls transaction flow before consensus, denial of whose service hurts the health and security of blockchain networks. This paper presents MPFUZZ, the first mempool fuzzer to find asymmetric DoS bugs by exploring the space of symbolized mempool states and optimistically estimating the promisingness of an intermediate state in reaching bug oracles. Compared to the baseline blockchain fuzzers, MPFUZZ achieves a > 100x speedup in finding known DETER exploits. Running MPFUZZ on major Ethereum clients leads to discovering new mempool vulnerabilities, which exhibit a wide variety of sophisticated patterns, including stealthy mempool eviction and mempool locking. Rule-based mitigation schemes are proposed against all newly discovered vulnerabilities.

cs.CR↗

arXiv2025

Response Uncertainty and Probe Modeling: Two Sides of the Same Coin in LLM Interpretability?

Probing techniques have shown promise in revealing how LLMs encode human-interpretable concepts, particularly when applied to curated datasets. However, the factors governing a dataset's suitability for effective probe training are not well-understood. This study hypothesizes that probe performance on such datasets reflects characteristics of both the LLM's generated responses and its internal feature space. Through quantitative analysis of probe performance and LLM response uncertainty across a series of tasks, we find a strong correlation: improved probe performance consistently corresponds to a reduction in response uncertainty, and vice versa. Subsequently, we delve deeper into this correlation through the lens of feature importance analysis. Our findings indicate that high LLM response variance is associated with a larger set of important features, which poses a greater challenge for probe models and often results in diminished performance. Moreover, leveraging the insights from response uncertainty analysis, we are able to identify concrete examples where LLM representations align with human knowledge across diverse domains, offering additional evidence of interpretable reasoning in LLMs.

cs.AI↗

arXiv2025

R1-ShareVL: Incentivizing Reasoning Capability of Multimodal Large Language Models via Share-GRPO

In this work, we aim to incentivize the reasoning ability of Multimodal Large Language Models (MLLMs) via reinforcement learning (RL) and develop an effective approach that mitigates the sparse reward and advantage vanishing issues during RL. To this end, we propose Share-GRPO, a novel RL approach that tackle these issues by exploring and sharing diverse reasoning trajectories over expanded question space. Specifically, Share-GRPO first expands the question space for a given question via data transformation techniques, and then encourages MLLM to effectively explore diverse reasoning trajectories over the expanded question space and shares the discovered reasoning trajectories across the expanded questions during RL. In addition, Share-GRPO also shares reward information during advantage computation, which estimates solution advantages hierarchically across and within question variants, allowing more accurate estimation of relative advantages and improving the stability of policy training. Extensive evaluations over six widely-used reasoning benchmarks showcase the superior performance of our method. Code will be available at https://github.com/HJYao00/R1-ShareVL.

cs.CV↗

arXiv2025

Ada-R1: Hybrid-CoT via Bi-Level Adaptive Reasoning Optimization

Recently, long-thought reasoning models achieve strong performance on complex reasoning tasks, but often incur substantial inference overhead, making efficiency a critical concern. Our empirical analysis reveals that the benefit of using Long-CoT varies across problems: while some problems require elaborate reasoning, others show no improvement, or even degraded accuracy. This motivates adaptive reasoning strategies that tailor reasoning depth to the input. However, prior work primarily reduces redundancy within long reasoning paths, limiting exploration of more efficient strategies beyond the Long-CoT paradigm. To address this, we propose a novel two-stage framework for adaptive and efficient reasoning. First, we construct a hybrid reasoning model by merging long and short CoT models to enable diverse reasoning styles. Second, we apply bi-level preference training to guide the model to select suitable reasoning styles (group-level), and prefer concise and correct reasoning within each style group (instance-level). Experiments demonstrate that our method (Ada-R1) significantly reduces inference costs compared to other baseline approaches, while maintaining performance. Notably, on five mathematical datasets, the average length of reasoning is reduced by more than 50%, highlighting the potential of adaptive strategies to optimize reasoning efficiency in large language models. Our code is coming soon at https://github.com/StarDewXXX/AdaR1

cs.AI↗

arXiv2025

Dimensionality-dependent electronic and vibrational dynamics in low-dimensional organic-inorganic tin halides

Photo-induced dynamics of electronic processes in materials are driven by the coupling between electronic and nuclear degrees of freedom. Here we construct 1D and 2D organic-inorganic tin halides to investigate the functional role of dimensionality to exciton-phonon coupling (EPC) and exciton self-trapping. The results show that the 1D system has strong EPC leading to excitation-independent self-trapped exciton (STE) emission, while the 2D system exhibits over ten times weaker EPC resulting in free exciton emission. By performing femtosecond transient absorption experiments, we directly resolve the room-temperature vibrational wavepackets in the 1D system, some of which propagate along the STE potential energy surface. A combination of wagging and asymmetric stretching motions (~106 cm-1) in tin iodide is identified as such a mode inducing exciton self-trapping. While no room-temperature wavepackets are observed in the 2D system. These findings uncover the interplay between the dimensionality-dependent EPC and electronic/nuclear dynamics, offering constructive guidance to develop multifunctional organic-inorganic metal halides.

physics.chem-ph↗

arXiv2025

IRNN: Innovation-driven Recurrent Neural Network for Time-Series Data Modeling and Prediction

Many real-world datasets are time series that are sequentially collected and contain rich temporal information. Thus, a common interest in practice is to capture dynamics of time series and predict their future evolutions. To this end, the recurrent neural network (RNN) has been a prevalent and effective machine learning option, which admits a nonlinear state-space model representation. Motivated by the resemblance between RNN and Kalman filter (KF) for linear state-space models, we propose in this paper Innovation-driven RNN (IRNN), a novel RNN architecture tailored to time-series data modeling and prediction tasks. By adapting the concept of "innovation" from KF to RNN, past prediction errors are adopted as additional input signals to update hidden states of RNN and boost prediction performance. Since innovation data depend on network parameters, existing training algorithms for RNN do not apply to IRNN straightforwardly. Thus, a tailored training algorithm dubbed input updating-based back-propagation through time (IU-BPTT) is further proposed, which alternates between updating innovations and optimizing network parameters via gradient descent. Experiments on real-world benchmark datasets show that the integration of innovations into various forms of RNN leads to remarkably improved prediction accuracy of IRNN without increasing the training cost substantially.

cs.LG↗

arXiv2025

Ultrafast dynamics of ferroelectric polarization of NbOI$_{2}$ captured with femtosecond electron diffraction

Two-dimensional (2D) ferroelectric materials like NbOI$_{2}$ have garnered significant interest, yet their temporal response and synergetic interaction with light remain underexplored. Previous studies on the polarization of oxide ferroelectrics have relied on time-resolved optical second harmonic generation or ultrafast X-ray scattering. Here, we probe the laser-induced polarization dynamics of 2D NbOI$_{2}$ nanocrystals using ultrafast transmission electron diffraction and deflectometry. The deflection of the electron pulses is directly sensitive to the changes in the polarization, while the diffraction signal captures the structural evolution. Excited with a UV laser pulse, the polarization of NbOI$_{2}$ is initially suppressed for two picoseconds, then it recovers and overshoots, leading to a transiently enhanced polarization persisting for over 200 ps. This recovery coincides with coherent acoustic phonon generation, triggering a piezoresponse in the NbOI$_{2}$ nanocrystals. Our results offer a new method for sensing the ferroelectric order parameter in femtosecond time scales.

cond-mat.mtrl-sci↗

arXiv2025

Advancing Tool-Augmented Large Language Models: Integrating Insights from Errors in Inference Trees

Tool-augmented large language models (LLMs) leverage tools, often in the form of APIs, to improve their reasoning capabilities on complex tasks. This enables them to act as intelligent agents interacting with the real world. The recently introduced ToolLLaMA model by Qin et al. [2023] utilizes the depth-first search-based decision tree (DFSDT) mechanism for multi-step reasoning with $16000+$ real-world APIs, effectively enhancing the performance of tool-augmented LLMs compared to traditional chain reasoning mechanisms. However, their approach only employs successful paths from decision trees (also called inference trees) for supervised fine-tuning (SFT), missing out on the potential learning opportunities from failed paths. Inspired by this, we propose an inference trajectory optimization framework based on preference learning to address this limitation. We first introduce a novel method for constructing step-wise preference data from tree-like expert trajectories, which leverages the previously ignored failed explorations in the decision trees. In the subsequent training phase, we first fine-tune the LLM with successful tool-usage expert trajectories and then apply direct preference optimization (DPO) with the preference data to update the LLM's policy, resulting in our ToolPrefer-LLaMA (TP-LLaMA) model. This approach not only enhances the utilization of original expert data but also broadens the learning space of the model. Our experiments demonstrate that by obtaining insights from errors in inference trees, TP-LLaMA significantly outperforms the baselines across almost all test scenarios by a large margin and exhibits better generalization capabilities with unseen APIs. At the same time, TP-LLaMA has also demonstrated superior reasoning efficiency compared to the baselines, making it more suitable for complex tool-usage reasoning tasks.

cs.CL↗

arXiv2025

Uncertainty in Graph Neural Networks: A Survey

Graph Neural Networks (GNNs) have been extensively used in various real-world applications. However, the predictive uncertainty of GNNs stemming from diverse sources such as inherent randomness in data and model training errors can lead to unstable and erroneous predictions. Therefore, identifying, quantifying, and utilizing uncertainty are essential to enhance the performance of the model for the downstream tasks as well as the reliability of the GNN predictions. This survey aims to provide a comprehensive overview of the GNNs from the perspective of uncertainty with an emphasis on its integration in graph learning. We compare and summarize existing graph uncertainty theory and methods, alongside the corresponding downstream tasks. Thereby, we bridge the gap between theory and practice, meanwhile connecting different GNN communities. Moreover, our work provides valuable insights into promising directions in this field.

cs.LG↗

arXiv2025

WalnutData: A UAV Remote Sensing Dataset of Green Walnuts and Model Evaluation

The UAV technology is gradually maturing and can provide extremely powerful support for smart agriculture and precise monitoring. Currently, there is no dataset related to green walnuts in the field of agricultural computer vision. Thus, in order to promote the algorithm design in the field of agricultural computer vision, we used UAV to collect remote-sensing data from 8 walnut sample plots. Considering that green walnuts are subject to various lighting conditions and occlusion, we constructed a large-scale dataset with a higher-granularity of target features - WalnutData. This dataset contains a total of 30,240 images and 706,208 instances, and there are 4 target categories: being illuminated by frontal light and unoccluded (A1), being backlit and unoccluded (A2), being illuminated by frontal light and occluded (B1), and being backlit and occluded (B2). Subsequently, we evaluated many mainstream algorithms on WalnutData and used these evaluation results as the baseline standard. The dataset and all evaluation results can be obtained at https://github.com/1wuming/WalnutData.

cs.CV↗

arXiv2025

KGARevion: An AI Agent for Knowledge-Intensive Biomedical QA

Biomedical reasoning integrates structured, codified knowledge with tacit, experience-driven insights. Depending on the context, quantity, and nature of available evidence, researchers and clinicians use diverse strategies, including rule-based, prototype-based, and case-based reasoning. Effective medical AI models must handle this complexity while ensuring reliability and adaptability. We introduce KGARevion, a knowledge graph-based agent that answers knowledge-intensive questions. Upon receiving a query, KGARevion generates relevant triplets by leveraging the latent knowledge embedded in a large language model. It then verifies these triplets against a grounded knowledge graph, filtering out errors and retaining only accurate, contextually relevant information for the final answer. This multi-step process strengthens reasoning, adapts to different models of medical inference, and outperforms retrieval-augmented generation-based approaches that lack effective verification mechanisms. Evaluations on medical QA benchmarks show that KGARevion improves accuracy by over 5.2% over 15 models in handling complex medical queries. To further assess its effectiveness, we curated three new medical QA datasets with varying levels of semantic complexity, where KGARevion improved accuracy by 10.4%. The agent integrates with different LLMs and biomedical knowledge graphs for broad applicability across knowledge-intensive tasks. We evaluated KGARevion on AfriMed-QA, a newly introduced dataset focused on African healthcare, demonstrating its strong zero-shot generalization to underrepresented medical contexts.

cs.AI↗