arXiv subjects

Jun Zhang

Publications and source records attributed to Jun Zhang.

At least 181 records · Page 10Linked to original sources

arXiv2026

Efficient Inference for Large Vision-Language Models: Bottlenecks, Techniques, and Prospects

Large Vision-Language Models (LVLMs) enable sophisticated reasoning over images and videos, yet their inference is hindered by a systemic efficiency barrier known as visual token dominance. This overhead is driven by a multi-regime interplay between high-resolution feature extraction, quadratic attention scaling, and memory bandwidth constraints. We present a systematic taxonomy of efficiency techniques structured around the inference lifecycle, consisting of encoding, prefilling, and decoding. Unlike prior reviews focused on isolated optimizations, we analyze the end-to-end pipeline to reveal how upstream decisions dictate downstream bottlenecks, covering compute-bound visual encoding, the intensive prefilling of massive contexts, and the ''visual memory wall'' in bandwidth-bound decoding. By decoupling the efficiency landscape into the axes of shaping information density, managing long-context attention, and overcoming memory limits, this work provides a structured analysis of how isolated optimizations compose to navigate the trade-off between visual fidelity and system efficiency. The survey concludes by outlining four future frontiers supported by pilot empirical insights, including hybrid compression based on functional unit sensitivity, modality-aware decoding with relaxed verification, progressive state management for streaming continuity, and stage-disaggregated serving through hardware-algorithm co-design. Our literature repository is at https://github.com/SuDIS-ZJU/Efficient-LVLMs-Inference.

cs.CL↗

arXiv2026

AgentSociety: Large-Scale Simulation of LLM-Driven Generative Agents Advances Understanding of Human Behaviors and Society

Understanding human behavior and society is a central focus in social sciences, with the rise of generative social science marking a significant paradigmatic shift. By leveraging bottom-up simulations, it replaces costly and logistically challenging traditional experiments with scalable, replicable, and systematic computational approaches for studying complex social dynamics. Recent advances in large language models (LLMs) have further transformed this research paradigm, enabling the creation of human-like generative social agents and realistic simulacra of society. In this paper, we propose AgentSociety, a large-scale social simulator that integrates LLM-driven agents, a realistic societal environment, and a powerful large-scale simulation engine. Based on the proposed simulator, we generate social lives for over 10k agents, simulating their 5 million interactions both among agents and between agents and their environment. Furthermore, we explore the potential of AgentSociety as a testbed for computational social experiments, focusing on five key social issues: polarization, the spread of inflammatory messages, the effects of universal basic income policies, the impact of external shocks such as hurricanes, and urban sustainability. These five issues serve as valuable cases for assessing AgentSociety's support for typical research methods -- such as surveys, interviews, and interventions -- as well as for investigating the patterns, causes, and underlying mechanisms of social issues. The alignment between AgentSociety's outcomes and real-world experimental results not only demonstrates its ability to capture human behaviors and their underlying mechanisms, but also underscores its potential as an important platform for social scientists and policymakers.

cs.SI↗

arXiv2026

Continuous Quantum Aperture: Beamforming with a Single-Vapor-Cell Rydberg Receiver

Beamforming is conventionally understood as a collective property of many discrete antenna elements in both communication and radar fields, which links angular selectivity to array size, element spacing, and band-specific hardware. Here we uncover a fundamentally different beamforming mechanism achieved by a Rydberg atomic receiver: a Rydberg-atom vapor cell dressed by a local-oscillator field constitutes a continuous quantum aperture. In this regime, spatially-varying quantum coherence across the aperture provides continuous amplitude-phase control, allowing a directional beam pattern to emerge from one sensing volume rather than from an engineered array. We establish the theory of continuous quantum aperture and show that tailoring the local-oscillator field can directly program the aperture response. This enables reconfigurable single-peak, multipeak, and multiband beamforming within a single vapor cell. Experiments on a Rydberg atomic receiver prototype verify that practical beam patterns agree with theoretical predictions across aperture sizes, frequency bands, and local-oscillator configurations. Leveraging this new beamforming mechanism, we further demonstrate interference mitigation, multiuser access, and multiband multiuser access with the single-vapor-cell platform. Our results identify the continuous quantum aperture as a new operating principle of Rydberg atomic receivers and establish single-vapor-cell beamforming as an integrated and reconfigurable platform for spatially selective electromagnetic reception.

quant-ph↗

arXiv2026

See the Forest for the Trees: Loosely Speculative Decoding via Visual-Semantic Guidance for Efficient Inference of Video LLMs

Video Large Language Models (Video-LLMs) excel in video understanding but suffer from high inference latency during autoregressive generation. Speculative Decoding (SD) mitigates this by applying a draft-and-verify paradigm, yet existing methods are constrained by rigid exact-match rules, severely limiting the acceleration potential. To bridge this gap, we propose LVSpec, the first training-free loosely SD framework tailored for Video-LLMs. Grounded in the insight that generation is governed by sparse visual-relevant anchors (mandating strictness) amidst abundant visual-irrelevant fillers (permitting loose verification), LVSpec employs a lightweight visual-relevant token identification scheme to accurately pinpoint the former. To further maximize acceptance, we augment this with a position-shift tolerant mechanism that effectively salvages positionally mismatched but semantically equivalent tokens. Experiments demonstrate that LVSpec achieves high fidelity and speed: it preserves >99.8 of target performance while accelerating Qwen2.5-VL-32B by 2.70x and LLaVA-OneVision-72B by 2.94x. Notably, it boosts the mean accepted length and speedup ratio by 136% and 35% compared to SOTA training-free SD methods for Video-LLMs.

cs.CL↗

arXiv2026

On the Step Length Confounding in LLM Reasoning Data Selection

Large reasoning models have recently demonstrated strong performance on complex tasks that require long chain-of-thought reasoning, through supervised fine-tuning on large-scale and high-quality datasets. To construct such datasets, existing pipelines generate long reasoning data from more capable Large Language Models (LLMs) and apply manually heuristic or naturalness-based selection methods to filter high-quality samples. Despite the proven effectiveness of naturalness-based data selection, which ranks data by the average log probability assigned by LLMs, our analysis shows that, when applied to LLM reasoning datasets, it systematically prefers samples with longer reasoning steps (i.e., more tokens per step) rather than higher-quality ones, a phenomenon we term step length confounding. Through quantitative analysis, we attribute this phenomenon to low-probability first tokens in reasoning steps; longer steps dilute their influence, thereby inflating the average log probabilities. To address this issue, we propose two variant methods: ASLEC-DROP, which drops first-token probabilities when computing average log probability, and ASLEC-CASL, which applies a causal debiasing regression to remove the first tokens' confounding effect. Experiments across four LLMs and five evaluation benchmarks demonstrate the effectiveness of our approach in mitigating the step length confounding problem.

cs.CL↗

arXiv2026

TC-AE: Unlocking Token Capacity for Deep Compression Autoencoders

We propose TC-AE, a ViT-based architecture for deep compression autoencoders. Existing methods commonly increase the channel number of latent representations to maintain reconstruction quality under high compression ratios. However, this strategy often leads to latent representation collapse, which degrades generative performance. Instead of relying on increasingly complex architectures or multi-stage training schemes, TC-AE addresses this challenge from the perspective of the token space, the key bridge between pixels and image latents, through two complementary innovations: Firstly, we study token number scaling by adjusting the patch size in ViT under a fixed latent budget, and identify aggressive token-to-latent compression as the key factor that limits effective scaling. To address this issue, we decompose token-to-latent compression into two stages, reducing structural information loss and enabling effective token number scaling for generation. Secondly, to further mitigate latent representation collapse, we enhance the semantic structure of image tokens via joint self-supervised training, leading to more generative-friendly latents. With these designs, TC-AE achieves substantially improved reconstruction and generative performance under deep compression. We hope our research will advance ViT-based tokenizer for visual generation.

cs.CV↗

arXiv2026

HybridKV: Hybrid KV Cache Compression for Efficient Multimodal Large Language Model Inference

Multimodal Large Language Models (MLLMs) have advanced unified reasoning over text, images, and videos, but their inference is hindered by the rapid growth of key-value (KV) caches. Each visual input expands into thousands of tokens, causing caches to scale linearly with context length and remain resident in GPU memory throughout decoding, which leads to prohibitive memory overhead and latency even on high-end GPUs. A common solution is to compress caches under a fixed allocated budget at different granularities: token-level uniformly discards less important tokens, layer-level varies retention across layers, and head-level redistributes budgets across heads. Yet these approaches stop at allocation and overlook the heterogeneous behaviors of attention heads that require distinct compression strategies. We propose HybridKV, a hybrid KV cache compression framework that integrates complementary strategies in three stages: heads are first classified into static or dynamic types using text-centric attention; then a top-down budget allocation scheme hierarchically assigns KV budgets; finally, static heads are compressed by text-prior pruning and dynamic heads by chunk-wise retrieval. Experiments on 11 multimodal benchmarks with Qwen2.5-VL-7B show that HybridKV reduces KV cache memory by up to $7.9\times$ and achieves $1.52\times$ faster decoding, with almost no performance drop or even higher relative to the full-cache MLLM.

cs.AI↗

arXiv2026

Training-Free Image Editing with Visual Context Integration and Concept Alignment

In image editing, it is essential to incorporate a context image to convey the user's precise requirements, such as subject appearance or image style. Existing training-based visual context-aware editing methods incur data collection effort and training cost. On the other hand, the training-free alternatives are typically established on diffusion inversion, which struggles with consistency and flexibility. In this work, we propose VicoEdit, a training-free and inversion-free method to inject the visual context into the pretrained text-prompted editing model. More specifically, VicoEdit directly transforms the source image into the target one based on the visual context, thereby eliminating the need for inversion that can lead to deviated trajectories. Moreover, we design a posterior sampling approach guided by concept alignment to enhance the editing consistency. Empirical results demonstrate that our training-free method achieves even better editing performance than the state-of-the-art training-based models.

cs.CV↗

arXiv2026

Tencent Advertising Algorithm Challenge 2025: All-Modality Generative Recommendation

Generative recommender systems are rapidly emerging as a new paradigm for recommendation, where collaborative identifiers and/or multi-modal content are mapped into discrete token spaces and user behavior is modelled with autoregressive sequence models. Despite progress on multi-modal recommendation datasets, there is still a lack of public benchmarks that jointly offer large-scale, realistic and fully all-modality data designed specifically for generative recommendation (GR) in industrial advertising. To foster research in this direction, we organised the Tencent Advertising Algorithm Challenge 2025, a global competition built on top of two all-modality datasets for GR: TencentGR-1M and TencentGR-10M. Both datasets are constructed from real de-identified Tencent Ads logs and contain rich collaborative IDs and multi-modal representations extracted with state-of-the-art embedding models. The preliminary track (TencentGR-1M) provides 1 million user sequences with up to 100 interacted items each, where each interaction is labeled with exposure and click signals, while the final track (TencentGR-10M) scales this to 10 million users and explicitly distinguishes between click and conversion events at both the sequence and target level. This paper presents the task definition, data construction process, feature schema, baseline GR model, evaluation protocol, and key findings from top-ranked and award-winning solutions. Our datasets focus on multi-modal sequence generation in an advertising setting and introduce weighted evaluation for high-value conversion events. We release our datasets at https://huggingface.co/datasets/TAAC2025 and baseline implementations at https://github.com/TencentAdvertisingAlgorithmCompetition/baseline_2025 to enable future research on all-modality generative recommendation at an industrial scale. The official website is https://algo.qq.com/2025.

cs.IR↗

arXiv2026

A Quantum-Driven Evolutionary Framework for Solving High-Dimensional Sharpe Ratio Portfolio Optimization

High-dimensional portfolio optimization faces significant computational challenges under complex constraints, with traditional optimization methods struggling to balance convergence speed and global exploration capability. To address this, firstly, we introduce an enhanced Sharpe ratio-based model that incorporates all constraints into the objective function using adaptive penalty terms, transforming the original constrained problem into an unconstrained single-objective formulation. This approach preserves financial interpretability while simplifying algorithmic implementation. To efficiently solve the resulting high-dimensional optimization problem, we develop a Quantum Hybrid Differential Evolution (QHDE) algorithm, which introduces a dynamic quantum tunneling mechanism that enables individuals to probabilistically escape local optima, dramatically enhancing global exploration and solution flexibility. To further improve performance, a good point set-chaos reverse learning strategy generates a well-dispersed initial population, providing a robust and diverse starting point. Meanwhile, a dynamic elite pool combined with Cauchy-Gaussian hybrid perturbations maintains population diversity and mitigates premature convergence, ensuring stable and high-quality solutions. Experimental validation on CEC benchmarks and real-world portfolios involving 20 to 80 assets demonstrates that QHDE's performance improves by up to 96.6%. It attains faster convergence, higher solution precision, and greater robustness than seven state-of-the-art counterparts, thereby confirming its suitability for complex, high-dimensional portfolio optimization.

cs.NE↗

arXiv2026

Boosting Document Parsing Efficiency and Performance with Coarse-to-Fine Visual Processing

Document parsing is a fine-grained task where image resolution significantly impacts performance. While advanced research leveraging vision-language models benefits from high-resolution input to boost model performance, this often leads to a quadratic increase in the number of vision tokens and significantly raises computational costs. We attribute this inefficiency to substantial visual regions redundancy in document images, like background. To tackle this, we propose PaddleOCR-VL, a novel coarse-to-fine architecture that focuses on semantically relevant regions while suppressing redundant ones, thereby improving both efficiency and performance. Specifically, we introduce a lightweight Valid Region Focus Module (VRFM) which leverages localization and contextual relationship prediction capabilities to identify valid vision tokens. Subsequently, we design and train a compact yet powerful 0.9B vision-language model (PaddleOCR-VL-0.9B) to perform detailed recognition, guided by VRFM outputs to avoid direct processing of the entire large image. Extensive experiments demonstrate that PaddleOCR-VL achieves state-of-the-art performance in both page-level parsing and element-level recognition. It significantly outperforms existing solutions, exhibits strong competitiveness against top-tier VLMs, and delivers fast inference while utilizing substantially fewer vision tokens and parameters, highlighting the effectiveness of targeted coarse-to-fine parsing for accurate and efficient document understanding. The source code and models are publicly available at https://github.com/PaddlePaddle/PaddleOCR.

cs.CV↗

arXiv2026

Geometry and dynamics on Liouville domains in $T^*\mathbb T^2$

Parallel to the study of toric domains, symplectically convex, and dynamically convex domains in $(\mathbb R^4, ω_{\rm std})$, we build an analogous framework and corresponding subclasses for Liouville domains in $(T^*\mathbb T^2,ω_{\rm can})$. A key feature of this framework is the introduction of a new notion of convexity, based on systolic ratios. Via various machinery in quantitative symplectic geometry, including ECH capacities, shape invariant, dynamical zeta function, etc., we investigate the relations between subclasses of Liouville domains in $T^*\mathbb T^2$, obtain large-scale geometry of Liouville domains in $T^*\mathbb T^2$ with respect to coarse Banach-Mazur distance, provide a non-flat codisc bundle of torus even under the action of exact symplectomorphisms, and verify the agreement of normalized capacities for a wide class of Liouville domains in $T^*\mathbb T^2$.

math.SG↗

arXiv2026

LGFNet: Local-Global Fusion Network with Fidelity Gap Delta Learning for Multi-Source Aerodynamics

The precise fusion of computational fluid dynamic (CFD) data, wind tunnel tests data, and flight tests data in aerodynamic area is essential for obtaining comprehensive knowledge of both localized flow structures and global aerodynamic trends across the entire flight envelope. However, existing methodologies often struggle to balance high-resolution local fidelity with wide-range global dependency, leading to either a loss of sharp discontinuities or an inability to capture long-range topological correlations. We propose Local-Global Fusion Network (LGFNet) for multi-scale feature decomposition to extract this dual-natured aerodynamic knowledge. To this end, LGFNet combines a spatial perception layer that integrates a sliding window mechanism with a relational reasoning layer based on self-attention, simultaneously reinforcing the continuity of fine-grained local features (e.g., shock waves) and capturing long-range flow information. Furthermore, the fidelity gap delta learning (FGDL) strategy is proposed to treat CFD data as a "low-frequency carrier" to explicitly approximate nonlinear discrepancies. This approach prevents unphysical smoothing while inheriting the foundational physical trends from the simulation baseline. Experiments demonstrate that LGFNet achieves state-of-the-art (SOTA) performance in both accuracy and uncertainty reduction across diverse aerodynamic scenarios.

cs.LG↗

arXiv2026

Physics-Informed Evolution: An Evolutionary Framework for Solving Quantum Control Problems Involving the Schrödinger Equation

Physics-informed Neural Networks (PINNs) show that embedding physical laws directly into the learning objective can significantly enhance the efficiency and physical consistency of neural network solutions. Similar to optimizing loss functions in machine learning, evolutionary algorithms iteratively optimize objective functions by simulating natural selection processes. Inspired by this principle, we ask a natural question: can physical information be similarly embedded into the fitness function of evolutionary algorithms? In this work, we propose Physics-informed Evolution (PIE), a novel framework that incorporates physical information derived from governing physical laws into the evolutionary fitness landscape, thereby extending Physics-informed artificial intelligence methods from machine learning to the broader domain of evolutionary computation. As a concrete instantiation, we apply PIE to quantum control problems governed by the Schrödinger equation, where the goal is to find optimal control fields that drive quantum systems from initial states to desired target states. We validate PIE on three representative quantum control benchmarks: state preparation in V-type three-level systems, entangled state generation in superconducting quantum circuits, and two-atom cavity QED systems. Within the PIE framework, we systematically compare the performance of ten single-objective and five multi-objective evolutionary algorithms. Experimental results demonstrate that by embedding physical information into the fitness function, PIE effectively guides evolutionary search, yielding control fields with high fidelity, low state deviation, and robust performance across different scenarios. Our findings further suggest that the Physics-informed principle extends naturally beyond neural network training to the broader domain of evolutionary computation.

quant-ph↗

arXiv2026

TimeLens: Rethinking Video Temporal Grounding with Multimodal LLMs

This paper does not introduce a novel method but instead establishes a straightforward, incremental, yet essential baseline for video temporal grounding (VTG), a core capability in video understanding. While multimodal large language models (MLLMs) excel at various video understanding tasks, the recipes for optimizing them for VTG remain under-explored. In this paper, we present TimeLens, a systematic investigation into building MLLMs with strong VTG ability, along two primary dimensions: data quality and algorithmic design. We first expose critical quality issues in existing VTG benchmarks and introduce TimeLens-Bench, comprising meticulously re-annotated versions of three popular benchmarks with strict quality criteria. Our analysis reveals dramatic model re-rankings compared to legacy benchmarks, confirming the unreliability of prior evaluation standards. We also address noisy training data through an automated re-annotation pipeline, yielding TimeLens-100K, a large-scale, high-quality training dataset. Building on our data foundation, we conduct in-depth explorations of algorithmic design principles, yielding a series of meaningful insights and effective yet efficient practices. These include interleaved textual encoding for time representation, a thinking-free reinforcement learning with verifiable rewards (RLVR) approach as the training paradigm, and carefully designed recipes for RLVR training. These efforts culminate in TimeLens models, a family of MLLMs with state-of-the-art VTG performance among open-source models and even surpass proprietary models such as GPT-5 and Gemini-2.5-Flash. All codes, data, and models will be released to facilitate future research.

cs.CV↗

arXiv2026

ByteLoom: Weaving Geometry-Consistent Human-Object Interactions through Progressive Curriculum Learning

Human-object interaction (HOI) video generation has garnered increasing attention due to its promising applications in digital humans, e-commerce, advertising, and robotics imitation learning. However, existing methods face two critical limitations: (1) a lack of effective mechanisms to inject multi-view information of the object into the model, leading to poor cross-view consistency, and (2) heavy reliance on fine-grained hand mesh annotations for modeling interaction occlusions. To address these challenges, we introduce ByteLoom, a Diffusion Transformer (DiT)-based framework that generates realistic HOI videos with geometrically consistent object illustration, using simplified human conditioning and 3D object inputs. We first propose an RCM-cache mechanism that leverages Relative Coordinate Maps (RCM) as a universal representation to maintain object's geometry consistency and precisely control 6-DoF object transformations in the meantime. To compensate HOI dataset scarcity and leverage existing datasets, we further design a training curriculum that enhances model capabilities in a progressive style and relaxes the demand of hand mesh. Extensive experiments demonstrate that our method faithfully preserves human identity and the object's multi-view geometry, while maintaining smooth motion and object manipulation.

cs.CV↗

arXiv2026

PP-OCRv5: A Specialized 5M-Parameter Model Rivaling Billion-Parameter Vision-Language Models on OCR Tasks

The advent of "OCR 2.0" and large-scale vision-language models (VLMs) has set new benchmarks in text recognition. However, these unified architectures often come with significant computational demands, challenges in precise text localization within complex layouts, and a propensity for textual hallucinations. Revisiting the prevailing notion that model scale is the sole path to high accuracy, this paper introduces PP-OCRv5, a meticulously optimized, lightweight OCR system with merely 5 million parameters. We demonstrate that PP-OCRv5 achieves performance competitive with many billion-parameter VLMs on standard OCR benchmarks, while offering superior localization precision and reduced hallucinations. The cornerstone of our success lies not in architectural expansion but in a data-centric investigation. We systematically dissect the role of training data by quantifying three critical dimensions: data difficulty, data accuracy, and data diversity. Our extensive experiments reveal that with a sufficient volume of high-quality, accurately labeled, and diverse data, the performance ceiling for traditional, efficient two-stage OCR pipelines is far higher than commonly assumed. This work provides compelling evidence for the viability of lightweight, specialized models in the large-model era and offers practical insights into data curation for OCR. The source code and models are publicly available at https://github.com/PaddlePaddle/PaddleOCR.

cs.CV↗

arXiv2026

WirelessBench: A Tolerance-Aware LLM Agent Benchmark for Wireless Network Intelligence

LLM agents are emerging as a key enabler for autonomous wireless network management. Reliably deploying them, however, demands benchmarks that reflect real engineering risk. Existing wireless benchmarks evaluate single isolated capabilities and treat all errors uniformly, missing both cascaded-chain failures and catastrophic unit confusions (\textit{e.g.}, dB vs.\ dBm). We present \wb{}, the first tolerance-aware, tool-integrated benchmark for LLM-based wireless agents. \wb{} is organized as a three-tier cognitive hierarchy: domain knowledge reasoning (WCHW, 1{,}392 items), intent-driven resource allocation (WCNS, 1{,}000 items), and proactive multi-step decisions under mobility (WCMSA, 1{,}000 items). Moreover, \wb{} is established on three design principles: \emph{(i)}~tolerance-aware scoring with catastrophic-error detection; \emph{(ii)}~tool-necessary tasks requiring a 3GPP-compliant ray-tracing query for channel quality; and \emph{(iii)}~Chain-of-Thought (CoT)-traceable items, where every benchmark item ships with a complete CoT trajectory enabling fine-grained diagnosis of where in the reasoning chain an agent fails. Our numerical results show that the direct-prompting model (GPT-4o) scores $68\%$, trailing a tool-integrated agent ($84.64\%$) by $16.64$\,pp; $23\%$ of errors are catastrophic failures invisible to exact-match metrics. More importantly, the hierarchy decomposes errors into four actionable diagnostic categories that flat evaluation cannot reveal. Code and data: https://wirelessbench.github.io/.

cs.NI↗