探索全球前沿学术脉络

01.

arXiv (CS.AI) 2026-06-24 DOI: arXiv:2606.23690

Beyond the Autoregressive Horizon: A Comprehensive Survey of Diffusion Models, World Modelling, and State Space Models for Code

作者:

Kishan Maharaj↗Ashita Saxena↗Srikanth Tamilselvam↗

arXiv:2606.23690v1 Announce Type: cross Abstract: Autoregressive (AR) language models have driven significant progress in automated software engineering, enabling powerful code generation and assistance systems. However, the next-token prediction paradigm introduces structural limitations for code reasoning, including restricted global planning, challenges in maintaining long-range dependencies, and limited grounding in program execution semantics. Noting the heavy skewness of existing literature towards AR models, we discuss emerging paradigms that could potentially overcome the logic and scaling bottlenecks of next-token prediction by unlocking next-generation architectural capabilities for code intelligence. Specifically, we discuss the potential of Diffusion Models, which generate code via holistic denoising that captures long-range syntactic constraints often missed by AR models. We also discuss Code World Models (CWMs), which simulate execution states to support reasoning, and State Space Models (SSMs), which provide linear-time efficiency for massive contexts. By connecting these developments with findings from cognitive neuroscience, we outline directions for developing "System 2" code generation agents.

阅读与讨论 → 访问原文 →

02.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2602.15720

ToaSt: Token Channel Selection and Structured Pruning for Efficient ViT

作者:

Hyunchan Moon↗Cheonjun Park↗Steven L. Waslander↗

Vision Transformers (ViTs) have achieved remarkable success across various vision tasks, yet their deployment is often hindered by prohibitive computational costs. While structured weight pruning and token compression have emerged as promising solutions, they suffer from prolonged retraining and inter-layer dependencies that complicate optimization, respectively. We propose ToaSt, a decoupled framework applying specialized strategies to distinct ViT components. We apply coupled head-wise structured pruning to Multi-Head Self-Attention modules, leveraging attention operation characteristics to enhance robustness. For Feed-Forward Networks (over 60% of FLOPs), we introduce Token Channel Selection (TCS), a training-free method that filters redundant noise channels at inference time. Extensive evaluations across nine diverse models, including DeiT, ViT-MAE, and Swin Transformer, demonstrate that ToaSt achieves superior trade-offs between accuracy and efficiency, consistently outperforming existing baselines. On ViT-MAE-Huge, ToaSt achieves 88.52% accuracy (+1.64%p) with 39.4% FLOPs reduction. ToaSt also transfers effectively to diverse downstream tasks (COCO detection, ADE20K segmentation, CIFAR-100 classification), achieving 52.2 versus 51.9 mAP on COCO. Code: github.com/SHANNonLab-HUFS/ToaSt

阅读与讨论 → 访问原文 →

03.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.00435

Detect Before You Leap: Mirage Detection in Vision-Language Models

作者:

Sayeed Shafayet Chowdhury↗Md. Shaown Miah↗S. M. Taiabul Haque↗Syed Ishtiaque Ahmed↗

Vision-language models (VLMs) can produce confident visual answers even when the required visual evidence is missing, blank, or unrelated to the question. This failure mode, recently described as mirage (mirage2026), is especially concerning in medical and document VQA, where a plausible but visually ungrounded answer may be mistaken for image-based evidence. We study the complementary problem of pre-release mirage detection: given an image-question pair, determine whether the VLM should answer or abstain before generation. To that end, we propose a novel model-agnostic Text-Conditioned Layer-wise Internal Alignment (TC-LIA) method that probes patch-token representations across the layers of a CLIP ViT-H/14 vision encoder. The key idea is to project layer-wise image patch tokens into the final CLIP embedding space and measure their similarity with the question embedding, thereby tracking whether question-relevant visual evidence emerges across vision layers. TC-LIA summarizes this alignment trajectory using final image-text cosine similarity, late-layer top-k patch-text alignment, early-to-late gain, and layer-wise slope. These features are combined with pixel-statistic based blank/noise detection, zero-shot domain routing, and structured VLM self-assessment in an ensemble. Across five VQA domains with related, unrelated-real, and blank/noise inputs, and across twelve VLM backbones, Qwen2.5-VL-32B achieves the highest three-class detection accuracy of 94.7% with a 3.0% mirage rate, while Qwen2.5-VL-72B achieves 94.6% accuracy with a lower 2.8% mirage rate. Baseline mirage rates span 21.7-66.6%.

阅读与讨论 → 访问原文 →

04.

arXiv (CS.CL) 2026-06-18 DOI: arXiv:2510.04120

Probing Semantic Alignment, Lexical Invariance, and Syntactic Influence in LLM Metaphor Processing

作者:

Fengying Ye↗Shanshan Wang↗Lidia S. Chao↗Derek F. Wong↗

Large language models (LLMs) achieve strong performance on metaphor detection and interpretation tasks, yet it remains unclear what such behavioral success reveals about metaphor processing. We present a diagnostic analysis that examines the limits of behavioral evidence by probing three complementary dimensions: semantic attribute alignment, lexical invariance, and syntactic sensitivity. Using geometric probing, we assess whether model-generated interpretations align with reference semantic attributes; through context-varying substitution, we analyze the stability of lexical associations between metaphorical and literal expressions; and via controlled syntactic perturbations, we examine sensitivity in metaphor detection. Our analysis reveals that LLM-generated interpretations can exhibit semantic drift relative to reference attributes; stable lexical anchors persist across contextual conditions, potentially supporting conventional metaphors while biasing novel metaphors requiring contextual integration; and detection performance is sensitive to syntactic irregularities. These findings suggest that strong behavioral performance may reflect heterogeneous underlying signals, highlighting the need for caution when interpreting metaphor benchmarks as evidence of robust, integrated semantic understanding.

阅读与讨论 → 访问原文 →

05.

arXiv (CS.AI) 2026-06-15 DOI: arXiv:2606.14218

Universal Manipulation Exoskeleton: Learning Compliant Whole-body Policies with Real-time Torque Feedback

作者:

Litian Liang↗Jingxi Xu↗Xinda Qi↗Yujun Cai↗Houzhu Ding↗Luqi Wang↗Zhixin Sun↗Jyh-Herng Chow↗Ming Yang↗Mark Cutkosky↗

arXiv:2606.14218v1 Announce Type: cross Abstract: For robots to work safely in household environments, they need to be compliant and react to torque and force feedback during contact. However, the majority of existing data collection pipelines still lack the ability to capture force and torque data for learning active compliant policies. In this paper, we present Universal Manipulation Exoskeleton (UME), an upper-limb exoskeleton that provides real-time haptic torque feedback while recording whole-arm configurations and joint torque signals for teleoperation. With transparent torque feedback, human operators can even unsheathe kinematically constrained objects while blindfolded. UME is low-cost, lightweight, and portable. Equipped with an embedded IMU, it enables teleoperation for mobile manipulation. With our proposed universal retargeting algorithm, UME can teleoperate a range of robots, including the 7DoF OpenArm, 7DoF Franka, and 6DoF X-ARM. We demonstrate that this combination of capabilities enables learning bimanual, whole-body, and active compliant policies that operate effectively in highly constrained spaces. The learned robust autonomous policies achieve high success rates across a variety of tasks, including long-horizon mobile manipulation, force-mediated box flipping, visually occluded box pushing, and space-constrained tabletop manipulation. Videos, code, and additional information can be found at https://ume-exo.github.io.

阅读与讨论 → 访问原文 →

06.

arXiv (CS.AI) 2026-06-24 DOI: arXiv:2606.24884

InSight: Self-Guided Skill Acquisition via Steerable VLAs

作者:

Maggie Wang↗Lars Osterberg↗Stephen Tian↗Ola Shorinwa↗Jiajun Wu↗Mac Schwager↗

arXiv:2606.24884v1 Announce Type: cross Abstract: Vision-language-action (VLA) models can learn manipulation skills from demonstrations, but their capabilities are bounded by the skills in the training data. We present InSight, a framework that unlocks autonomous skill acquisition by rendering VLAs steerable at the primitive-action level (e.g., "move gripper to the bowl", "lift upward", "pour the bottle"). InSight consists of two primary stages: (1) an automated segmentation pipeline that partitions demonstrations into labeled primitives via VLM plan decomposition and end-effector poses to enable VLA primitive steerability, and (2) a VLM-guided data flywheel that identifies missing primitives required to accomplish a novel task, autonomously attempts demonstrations of the missing primitives with VLM-proposed low-level control, and automatically labels, stores, and integrates successful demonstrations into the VLA training set. We evaluate InSight across simulation and real-world manipulation tasks, including block flipping, drawer closing, sweeping, twisting, and pouring, without any human demonstrations of these target skills. Once learned, these primitives can be composed to execute novel, long-horizon tasks without additional human demonstrations. Our findings demonstrate that primitive steerability provides a practical foundation for continual skill acquisition in VLA policies. Project website: https://insight-vla.github.io.

阅读与讨论 → 访问原文 →

07.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2606.14817

Combining Retrieval-Augmented Text Generation with LLMs for Reading Content Recommendations

作者:

Sooyeon Kim↗Piotr S. Maci\k{a}g↗

arXiv:2606.14817v1 Announce Type: cross Abstract: This work presents the design, implementation, and evaluation of a system for generating personalized reading content using Large Language Models (LLMs) combined with Retrieval-Augmented Generation (RAG). The proposed architecture consists of four modules: Input, RAG, Generation, and Judging and enables users to specify both a question and a target reading content complexity. RAG is employed to retrieve relevant information from the Internet, enriching and grounding the content produced by three modern LLMs: Meta LLaMA 4 Scout, LLaMA 3.1 8B Instant, and Google Gemma2 9B. Reading materials are generated using three prompting strategies (Chain-of-Thought, zero-shot, and few-shot), and the LLM-as-a-Judge module automatically evaluates answer quality and alignment with the desired readability level. Experimental results show that RAG consistently improves system performance across all models and prompting techniques, increasing relevance and particularly groundedness by up to 26-35 percentage points. Overall, the findings demonstrate that the RAG-augmented architecture effectively produces reading content tailored to user queries and desired textual complexity.

阅读与讨论 → 访问原文 →

08.

arXiv (CS.CL) 2026-06-18 DOI: arXiv:2606.18543

CEO-Bench: Can Agents Play the Long Game?

作者:

Haozhe Chen↗Karthik Narasimhan↗Zhuang Liu↗

Language model agents are becoming proficient executors at isolated, short-horizon tasks such as software engineering and customer service. Yet real-world challenges require a combination of sophisticated skills that remain largely untested in agents: (1) navigating long horizons amid uncertainty; (2) acquiring information in noisy environments; (3) adapting to a changing world; (4) orchestrating multiple moving parts toward a coherent goal. We introduce CEO-Bench, which evaluates these capabilities together by simulating a representative real-world task: operating a startup for 500 days. An agent manages pricing, marketing, budgeting, and many other aspects of a fictional company through a programmable Python interface, operating in the same environment and facing the same challenges as a human CEO. Success demands analyzing noisy, interconnected business databases, translating signals into sound strategy, and coordinating many decisions with programming. The strongest agents write sophisticated code that simulates customer cohorts to forecast future cash and mines negotiation history to uncover hidden customer preferences. Even so, most state-of-the-art models struggle in this environment. Only Claude Opus 4.8 and GPT-5.5 finish above the $1M starting balance, and neither consistently turns a profit. CEO-Bench takes a first step toward measuring the intelligence required to drive sustained, adaptive progress over time.

阅读与讨论 → 访问原文 →

09.

arXiv (CS.CL) 2026-06-18 DOI: arXiv:2606.18947

Decoupling Search from Reasoning: A Vendor-Agnostic Grounding Architecture for LLM Agents

作者:

Emmanuel Aboah Boateng↗Kyle MacDonald↗Amardeep Kumar↗Siddharth Kodwani↗Sudeep Das↗

Production LLM agents increasingly depend on real-time search, yet native search grounding bundles retrieval policy, provider choice, evidence injection, cost, latency, and generation behavior behind a single model-provider boundary. This coupling makes grounding hard to inspect, tune, reuse, or port, and can trigger Search-Induced Verbosity that breaks strict output contracts. We present Decoupled Search Grounding (DSG), a vendor-agnostic boundary that moves grounding outside the reasoning model through an MCP-compatible gateway, exposing provider routing, source-aware context rendering, configured fallback, retrieval-depth control, and exact plus semantic caching as first-class controls. Across five frontier models on SimpleQA, FreshQA, and HotpotQA, native search leads on recency-sensitive FreshQA, but DSG exposes a stronger frontier when control matters: on SimpleQA it nearly matches native accuracy (86.1% vs. 87.7%) at 91% lower search cost, preserves concise answer contracts, and reaches a 99.4% warm-cache hit rate with 68% lower latency. Deployed as a shared production grounding layer for large-scale agentic workloads with interchangeable models, DSG matches or slightly exceeds native-search accuracy on an e-commerce query-understanding (QIU) workload while cutting search cost by over 98%. Real-time grounding is best treated as an optimizable interface boundary, not a fixed model feature.

阅读与讨论 → 访问原文 →

10.

arXiv (CS.CV) 2026-06-19 DOI: arXiv:2606.20103

Geometry-Preserving in 3D Gaussian Splatting for LiDAR-Camera Extrinsic Calibration

作者:

Kyoleen Kwak↗Daeho Kim↗Jeong Woon Lee↗Hyoseok Hwang↗

Accurate LiDAR-camera calibration is essential for robust multi-modal perception. Targetless approaches avoid manual setup but remain limited by the scarcity of discriminative cross-modal features. Recent methods address this by reconstructing the scene within a differentiable model, enabling extrinsic optimization through dense photometric supervision. Among these, 3D Gaussian Splatting (3DGS) has been widely adopted as a geometric proxy that bridges LiDAR and camera within a single differentiable framework. However, since 3DGS was originally designed for novel view synthesis, existing methods tend to prioritize rendering quality, causing the proxy geometry to drift from the true LiDAR structure. We propose a framework that preserves the metric geometry of the Gaussian proxy by aggregating multi-view LiDAR observations for dense depth supervision and blocking photometric gradients from updating the Gaussian spatial parameters. We validate our method on public driving datasets, where it consistently outperforms existing targetless methods in calibration accuracy.

阅读与讨论 → 访问原文 →

11.

arXiv (CS.CV) 2026-06-18 DOI: arXiv:2511.20002

Semantic Router: On the Feasibility of Hijacking MLLMs via a Single Adversarial Perturbation

作者:

Changyue Li↗Jiaying Li↗Youliang Yuan↗Jiaming He↗Zhicong Huang↗Pinjia He↗

Multimodal Large Language Models (MLLMs) are increasingly deployed in stateless systems, such as autonomous driving and robotics. This paper investigates a novel threat: Semantic-Aware Hijacking. We explore the feasibility of hijacking multiple stateless decisions simultaneously using a single universal perturbation. We introduce the Semantic-Aware Universal Perturbation (SAUP), which acts as a semantic router, "actively" perceiving input semantics and routing them to distinct, attacker-defined targets. To achieve this, we conduct theoretical and empirical analysis on the geometric properties in the latent space. Guided by these insights, we propose the Semantic-Oriented (SORT) optimization strategy and annotate a new dataset with fine-grained semantics to evaluate performance. Extensive experiments on three representative MLLMs demonstrate the fundamental feasibility of this attack, achieving a 66% attack success rate over five targets using a single frame against Qwen.

阅读与讨论 → 访问原文 →

12.

arXiv (CS.AI) 2026-06-24 DOI: arXiv:2606.23983

Maestro Order: A Model-Agnostic Orchestration Harness

作者:

Hidayet Aksu↗

arXiv:2606.23983v1 Announce Type: cross Abstract: A single forward pass of a capable model is a fast, fluent, and unreliable problem-solver: it is right often enough to be useful and wrong often enough to be dangerous; in language models, such confident errors are known as hallucinations. We present Maestro Order, a model-agnostic orchestration harness that turns unreliable solvers into reliable problem-solving systems by composing them according to four structural primitives (decompose, ensemble, verify, and recurse) and a budget-aware controller that decides where to spend compute. The harness treats any model as a black-box base solver behind a uniform interface, layers a verifier ensemble whose discrimination is measured online, and allocates verification and voting to the stages with the highest marginal reliability per unit cost. We give the architecture, the message and state schema, the controller algorithm, and the engineering that makes it deterministic, observable, and fault-tolerant. We then specify an evaluation methodology (reliability at fixed cost, coverage, calibration, and ablations) and report results from a faithful Monte Carlo simulation of the harness over a parameterized solver/verifier model. The simulation reproduces the predicted laws quantitatively: verification amplifies reliability geometrically (e.g. $0.55\to0.98$ with two gates, $\to0.999$ with four), voting helps only above chance and is limited by shared errors, and a budget-aware controller reaches a target reliability at a small fraction of the cost of voting alone by selecting the cheapest mechanism for each regime. We close with failure modes (verifier gaming, correlated errors, and decomposition error compounding) and concrete guidance: build robust checkers, diversify solvers, and let the controller put compute where the information is.

阅读与讨论 → 访问原文 →

13.

arXiv (CS.LG) 2026-06-17 DOI: arXiv:2606.18166

Evaluating Open-Source LLMs for Multi-Label ATT&CK Technique Classification on CTI Reports

作者:

Ahmed Ryan↗Saad Sakib Noor↗Md Erfan↗Shaswata Mitra↗Sudip Mittal↗Md Rayhanur Rahman↗

arXiv:2606.18166v1 Announce Type: cross Abstract: Classifying Cyber Threat Intelligence (CTI) using MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) is essential for proactive defense, but historically required extensive human effort. Pre-Large Language Model (LLM) automation sped up this process, but could not resolve the complex language and multi-step attack patterns found in unstructured CTI reports. LLMs addressed previous limitations by using contextual reasoning to understand unstructured text. However, current evaluations rely on simplified, single-technique sentences that ignore the complexity of real-world CTI reports, which often leads to inflated performance results. Consequently, the baseline performance of open-source LLMs on complex unstructured CTI reports remains unevaluated. To address this gap, we constructed a ground-truth dataset of 2,076 human-annotated sentences (1,281 technique-positive, 795 negative) from 83 complex unstructured CTI reports. These sentences were mapped to 114 unique ATT&CK techniques using a six-phase annotation process, achieving \k{appa} = 0.68 inter-annotator agreement. Using this dataset, we evaluated seven open-source LLMs ranging from 8B to 236B parameters across prompt strategy and temperature configurations. The highest-performing LLM achieved a micro-averaged F1 score of 0.22, establishing the empirical baseline for multi-label ATT&CK classification on complex unstructured CTI. Parameter size showed a statistically significant positive correlation with F1 score. Prompt strategy and temperature produced no statistically significant gains across model configurations. These results indicate that current open-source LLMs are insufficient for production-grade ATT&CK classification. The dataset, benchmark, and findings provide a reproducible foundation for future CTI research.

阅读与讨论 → 访问原文 →

14.

arXiv (CS.LG) 2026-06-12 DOI: arXiv:2509.19526

Metriplectic Conditional Flow Matching for Dissipative Dynamics

作者:

Ali Baheri↗Lars Lindemann↗

arXiv:2509.19526v2 Announce Type: replace Abstract: Metriplectic conditional flow matching (MCFM) learns dissipative dynamics without violating first principles. Neural surrogates often inject energy and destabilize long-horizon rollouts; MCFM instead builds the conservative-dissipative split into both the vector field and a structure preserving sampler. MCFM trains via conditional flow matching on short transitions, avoiding long rollout adjoints. In inference, a Strang-prox scheme alternates a symplectic update with a proximal metric step, ensuring discrete energy decay; an optional projection enforces strict decay when a trusted energy is available. We provide continuous and discrete time guarantees linking this parameterization and sampler to conservation, monotonic dissipation, and stable rollouts. On a controlled mechanical benchmark, MCFM yields phase portraits closer to ground truth and markedly fewer energy-increase and positive energy rate events than an equally expressive unconstrained neural flow, while matching terminal distributional fit.

阅读与讨论 → 访问原文 →

15.

arXiv (CS.LG) 2026-06-12 DOI: arXiv:2606.12643

TEDD: Robust Detection of Unstable Temporal Features

作者:

Ricardo Ribeiro Pereira↗Bruno Casal Lara\~na↗N\'adia Soares↗Miguel Ara\'ujo↗

arXiv:2606.12643v1 Announce Type: new Abstract: When working with real-world temporal data, it is common to encounter features whose distribution is changing over time. The naive employment of Machine Learning models on this unstable data might lead to rapidly degrading performance, especially if the new distribution is much different from what was previously seen during training. In order to cope with this problem, it is critical to automatically identify features that are changing over time. With these features detected, data scientists and other practitioners will be able to mitigate the issue (for instance, by applying data transformations), deploying more robust models that retain high performance for longer periods of time. In this paper, we describe which temporal changes a feature should not suffer from, and propose TEDD, a technique to a) identify when a dataset might lead to an unstable Machine Learning model and b) automatically detect which features cause such lack of robustness. In order to achieve it, we leverage a regression model to highlight which features contribute to a good prediction of an instance's timestamp. We compare our approach to other methods in real and synthetic data, testing their detection capability on all simple change patterns. We show that our method: detects all types of basic changes, both for numerical and categorical features; can detect multivariate drifts; returns a comparable value measuring the amount of change of each feature; requires no parameter tuning; and is scalable both on number of features and instances of the dataset.

阅读与讨论 → 访问原文 →

16.

arXiv (quant-ph) 2026-06-12 DOI: arXiv:2505.23289

Intermediate State Formation of Topologically Associated Chromatin Domains using Quantum Annealing

作者:

Tobias Kempe↗S. M. Ali Tabei↗Mohammad H. Ansari↗

arXiv:2505.23289v2 Announce Type: replace Abstract: Topologically Associating Chromatin Domains are spatially distinct chromatin regions that regulate transcription by segregating active and inactive genomic elements. Empirical studies show that their formation correlates with local patterns of epigenetic markers, yet the precise mechanisms linking 1D epigenetic landscapes to 3D chromatin folding remain unclear. Recent models represent chromatin as a spin system, where nucleosomes are treated as discrete-state variables coupled by interaction strengths derived from genomic and epigenetic data. Classical samplers struggle with these models due to high frustration and dense couplings. Here, we present a quantum annealing (QA) approach to efficiently sample chromatin states, embedding an epigenetic Ising model into the topology of D-Wave quantum processors. Rather than reconstructing exact TAD size distributions or insulation scores, our method reproduces statistical features, such as mean marker incidences and intra-/inter-nucleosome correlations, while generating configurations that exhibit TAD-like structural motifs. These results demonstrate QA as an alternative to explore the chromatin architecture and provide a foundation in epigenetic modeling.

阅读与讨论 → 访问原文 →

17.

arXiv (CS.CV) 2026-06-17 DOI: arXiv:2606.17798

LiveStarPro: Proactive Streaming Video Understanding with Hierarchical Memory for Long-Horizon Streams

作者:

Zhenyu Yang↗Kairui Zhang↗Bing Wang↗Shengsheng Qian↗Changsheng Xu↗

Despite the remarkable progress of Video Large Language Models (Video-LLMs), current online architectures still struggle to simultaneously process continuous video streams, decide autonomously when to respond, and preserve long-horizon contextual memory. These obstacles undermine real-time responsiveness and cause severe forgetting throughout prolonged interactions. In this work, we introduce LiveStarPro, a live streaming assistant that is designed for proactive video understanding over long-horizon streams. The design of LiveStarPro rests on three complementary components. The first component is Streaming Verification Decoding (SVeD), an inference framework that identifies the appropriate response timing through single-pass perplexity verification, thereby eliminating the dependency on explicit silence tokens. The second component is Streaming Causal Attention Masks (SCAM), a training strategy that enforces incremental video-language alignment over variable-length streams. The third component is Tree-Structured Hierarchical Memory (TSHM), a recursive memory architecture that organizes evicted historical information into event chains and consequently enables efficient retrieval from effectively unbounded video streams. To facilitate a comprehensive evaluation under realistic online conditions, we further present OmniStarPro, a large-scale benchmark that spans 15 diverse real-world scenarios and that extends to hour-scale streams for the assessment of long-term recall. Extensive experiments demonstrate that LiveStarPro consistently surpasses existing methods, attaining a 28.9% improvement in semantic correctness and an 18.2% reduction in timing error, while its streaming key-value cache further yields a 1.58x inference speedup over the same model without caching. The model and the code are publicly available at https://github.com/sotayang/LiveStarPro.

阅读与讨论 → 访问原文 →

18.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.16067

Stepwise Token Selection for Efficient Multimodal Large Language Models

作者:

Landi He↗Shawn Young↗Lijian Xu↗

In multimodal large language models (MLLMs), inference cost is largely dominated by the visual token prefix rather than the language backbone, making token reduction a key factor for improving efficiency. Existing approaches typically assign independent importance scores to visual tokens and retain a fixed number of top-ranked tokens, implicitly assuming token independence and a uniform compression ratio across inputs. In this work, we reformulate visual token pruning as a sequential decision-making process. Specifically, we introduce a pointer-style selection mechanism that iteratively chooses informative tokens, conditioning each decision on previously selected ones, and dynamically determines when to stop via a learned termination action. This enables joint optimization of both the selected subset and its size. To enable end-to-end training under standard language modeling objectives, we design a differentiable relaxation based on a variance-preserving noise interpolation scheme, allowing gradients to propagate through the discrete selection process. Extensive experiments on LLaVA-v1.5-7B and Qwen2.5-VL-7B demonstrate that our approach consistently outperforms fixed-ratio baselines across different compression levels. Under aggressive pruning that removes 88.9% of visual tokens, our method preserves 94.6% of the original accuracy while achieving a 1.88x speed-up in prefill latency.

阅读与讨论 → 访问原文 →

19.

arXiv (CS.AI) 2026-06-18 DOI: arXiv:2606.19319

Data Intelligence Agents: Interpreting, Modeling, and Querying Enterprise Data via Autonomous Coding Agents

作者:

Anoushka Vyas↗Aarushi Dhanuka↗Sina Khoshfetrat Pakazad↗Henrik Ohlsson↗

arXiv:2606.19319v1 Announce Type: cross Abstract: Production data integration is bottlenecked by repeated, lossy handoffs between data owners, engineers, and analysts who must collaboratively discover, structure, and query enterprise data. We present Data Intelligence Agents (DIA), a system of three agents (Data Interpreter, Schema Creator, and Query Generator) that compresses this workflow by treating autonomous coding agents (ACAs) as a first-class abstraction: rather than emitting text, the agents generate, execute, validate, and repair concrete artifacts, draw on a shared memory for experience reuse, and surface each for review by domain experts. DIA is deployed in production for enterprise customers. We study the Query Generator in depth and evaluate it in fully autonomous mode across seven SQL benchmarks spanning four task categories and four dialects. It matches or surpasses the best published results on all seven, demonstrating that an architecture grounded in execution, built on ACAs and a shared memory, generalizes across the data intelligence workload with adaptation confined to natural-language instructions.

阅读与讨论 → 访问原文 →

20.

arXiv (CS.AI) 2026-06-24 DOI: arXiv:2511.20892

Representation Interventions Enable Lifelong Knowledge Memory Control in LLMs

作者:

Xuyuan Liu↗Shengyu Chen↗Xinshuai Dong↗Yanchi Liu↗Xujiang Zhao↗Haoyu Wang↗Yujun Yan↗Haifeng Chen↗Zhengzhang Chen↗

arXiv:2511.20892v4 Announce Type: replace Abstract: Large language models (LLMs) often produce incorrect or outdated content after being employed. Efficient and accurate knowledge updates without costly retraining are a major challenge. This problem is particularly challenging in lifelong settings, where complex, unstructured knowledge must coexist without interference. We introduce RILKE (Representation Intervention for Lifelong KnowledgE Control), a robust and scalable method that treats knowledge control as interventions within the model's representation space. Leveraging representation-space expressiveness, we identify two key properties enabling RILKE to achieve fine-grained control over complex, unstructured knowledge while maintaining general utility with frozen base weights. During training, RILKE learns paraphrase-robust and edit-localized modules that limit each update to a low-dimensional subspace to minimize cross-edit interference. At inference, a query-adaptive router selects the appropriate module to guide the model's generation. Across LLaMA and Qwen models, RILKE scales effectively to large-scale benchmarks, demonstrating high edit success and strong paraphrase generalization while preserving general utility with modest memory overhead. These results show RILKE is an effective and scalable solution for lifelong knowledge control in LLMs.

阅读与讨论 → 访问原文 →

21.

medRxiv (Medicine) 2026-06-23 DOI: HASH:b5f7a037a31b4cf9728c19c5d0d42190

Comparative Evaluation of Machine Learning and Deep Learning Models for Early Prediction of Severe Acute Pancreatitis: A Multi-Model Study Using the 2012 Revised Atlanta Classification

作者:

stern↗

**Background:** Acute pancreatitis (AP) is a common gastrointestinal emergency with a subset of patients progressing to severe acute pancreatitis (SAP), which carries substantial morbidity and mortality. Current clinical severity scores such as BISAP, APACHE II, Ranson, and the Modified CT Severity Index require upon 48 hours of observation before reliable assessment is possible, limiting early triage. Machine learning (ML) approaches using routine admission laboratory values may enable earlier, more accurate prediction. **Methods:** We evaluated 11 models spanning three architectural families classical ML (Logistic Regression, Random Forest, Gradient Boosting), feedforward deep learning (MLP, Residual MLP, Attention MLP), and recurrent deep learning (LSTM, Stacked LSTM, Bidirectional LSTM, LSTM+Attention, CNN-LSTM) on a Chinese AP cohort of 722 patients (585 severe, 137 mild) labelled according to the 2012 Revised Atlanta Classification. Performance was assessed via 5-fold stratified cross-validation using AUC-ROC, F1 score, sensitivity, specificity, and PPV, with decision thresholds optimised for maximal F1. **Results:** Random Forest achieved the highest AUC of 0.877 (F1=0.917, sensitivity=96.8%, PPV=87.1%), followed closely by Gradient Boosting (AUC=0.874, F1=0.918). Classical ML models consistently outperformed deep learning counterparts. CNN-LSTM was the best recurrent model (AUC=0.777) but remained inferior to all classical approaches. LSTM-family models produced AUC values of 0.684-0.777, reflecting the cross-sectional tabular nature of the data. **Conclusions:** Random Forest provides robust, high-sensitivity early prediction of SAP severity using routine admission data. External prospective validation is required before clinical deployment. **Keywords:** acute pancreatitis; severity prediction; machine learning; random forest; deep learning; LSTM; Revised Atlanta Classification; early triage

阅读与讨论 → 访问原文 →

22.

arXiv (CS.CV) 2026-06-24 DOI: arXiv:2604.00725

A Benchmark of State-Space Models vs. Transformers and BiLSTM-based Models for Historical Newspaper OCR

作者:

Merveilles Agbeti-Messan↗Pierrick Tranouez↗St\'ephane Nicolas↗Cl\'ement Chatelain↗Thierry Paquet↗

End-to-end OCR for historical newspapers remains challenging, as models must handle long text sequences, degraded print quality, and complex layouts. While Transformer-based recognizers dominate current research, their quadratic complexity limits efficient paragraph-level transcription and large-scale deployment. We investigate linear-time State-Space Models (SSMs), specifically Mamba, as a scalable alternative to Transformer-based sequence modeling for OCR. We present to our knowledge, the first OCR architecture based on SSMs, combining a CNN visual encoder with bi-directional and autoregressive Mamba sequence modeling, and conduct a large-scale benchmark comparing SSMs with Transformer- and BiLSTM-based recognizers. Multiple decoding strategies (CTC, autoregressive, and non-autoregressive) are evaluated under identical training conditions alongside strong neural baselines (VAN, DAN, DANIEL) and widely used off-the-shelf OCR engines (PERO-OCR, Tesseract OCR, TrOCR, Gemini). Experiments on historical newspapers from the Bibliotheque nationale du Luxembourg, with newly released >99% verified gold-standard annotations, and cross-dataset tests on Fraktur and Antiqua lines, show that all neural models achieve low error rates (~2% CER), making computational efficiency the main differentiator. Mamba-based models maintain competitive accuracy while halving inference time and exhibiting superior memory scaling (1.26x vs 2.30x growth at 1000 chars), reaching 6.07% CER at the severely degraded paragraph level compared to 5.24% for DAN, while remaining 2.05x faster. We release code, trained models, and standardized evaluation protocols to enable reproducible research and guide practitioners in large-scale cultural heritage OCR available at https://github.com/MarcoPerson/ssm-ocr-benchmark.

阅读与讨论 → 访问原文 →

23.

arXiv (CS.AI) 2026-06-18 DOI: arXiv:2606.18619

Code-Augur: Agentic Vulnerability Detection via Specification Inference

作者:

Zhengxiong Luo↗Mehtab Zafar↗Dylan Wolff↗Abhik Roychoudhury↗

arXiv:2606.18619v1 Announce Type: cross Abstract: The advent of agentic vulnerability detection is already becoming a watershed moment for software security. Audits conducted entirely by autonomous LLM agents are uncovering critical vulnerabilities in fundamental software underpinning digital society. Many of these vulnerabilities remained masked for years, surfacing only now with AI agents. Yet the reasoning behind these discoveries remains alarmingly opaque and unvalidated. What assumptions did the agent make about a function's inputs when it deemed that function to be secure? Failures in reasoning and incorrect assumptions can lead to missed vulnerabilities and reduce trust in agentic analysis. We propose a security-specification-first paradigm that (1) exposes the agent's tacit assumptions explicitly as security specifications and (2) continuously refines those specifications via runtime falsification. We realize our approach in Code-Augur, a novel harness for agentic vulnerability detection. Given a codebase, Code-Augur analyzes each component of the system for vulnerable code. When it deems a component to be secure, it commits the local invariants behind that judgment as in-source assertions. In parallel, Code-Augur leverages a guided fuzzer to attempt to falsify those assumptions. When the fuzzer triggers an assertion, this either reveals a genuine vulnerability or a flawed specification to refine. In both cases, this process grounds the agent's understanding, aligning its view of code intent with how the code actually behaves. On real-world subjects, Code-Augur effectively leverages security specifications to detect more vulnerabilities than other state-of-the-art agents. Additionally, Code-Augur found 22 new vulnerabilities in key open-source projects. Compared to curated specialized models like Claude Mythos, Code-Augur offers effective agentic vulnerability detection built on widely available LLMs like Sonnet and DeepSeek.

阅读与讨论 → 访问原文 →

24.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2606.15684

Multi-agent Framework for Time-Sensitive Complementary Collaboration in Minecraft

作者:

Juheon Yi↗Jinglu Wang↗Xiaoyi Zhang↗Yan Lu↗

arXiv:2606.15684v1 Announce Type: new Abstract: We present TickingCollabBench, a Minecraft-based multi-agent benchmark for a novel class of time-sensitive complementary collaboration tasks. Our benchmark reflects four core characteristics of real-world collaboration: agent heterogeneity, mandatory collaboration, dynamic environments, and strict real-time constraints with failure risks. To enable this, we develop the TickingCollab framework, which supports the generation of diverse dynamic environments and abstracts Minecraft's primitive APIs to enable declarative YAML task specifications for composing these events. Building on this, we design a feasibility-aware automated benchmark generation pipeline, where an LLM drafts structurally diverse task configurations and feasibility verifier filters out invalid ones using approximate constraints. Evaluations demonstrate that lang latency and inherent difficulty of coordinating under partial observability and agent heterogeneity cause LLMs to frequently fail under dynamic environments and fall significantly short of a global-knowledge oracle.

阅读与讨论 → 访问原文 →

25.

arXiv (CS.CV) 2026-06-17 DOI: arXiv:2606.17639

ERQA-Plus: A Diagnostic Benchmark for Reasoning in Embodied AI

作者:

Hong Yang↗Basura Fernando↗

Generalist embodied agents require more than object recognition: they must reason about spatial relations, actions, procedures, human intentions, environmental constraints, and commonsense consequences from situated visual observations. Yet existing visual and embodied question answering benchmarks often provide limited control over the reasoning dependencies being tested, making it difficult to distinguish grounded embodied reasoning from shortcut-driven visual or linguistic pattern matching. We present ERQA-Plus, a diagnostic benchmark for reasoning in embodied AI. ERQA-Plus contains 1,766 question-answer instances grounded in 711 robot-centric images and organized according to a structured taxonomy spanning perceptual, action-centric, social-interaction, navigation-environmental, and contextual commonsense reasoning. The dataset is constructed using a multi-stage generation and validation pipeline that combines taxonomy-guided question generation, automatic quality judging, iterative revision, and human assessment to improve visual grounding, answer validity, and reasoning quality. We benchmark representative general-purpose vision-language models and embodied models, including LLaVA-NeXT-8B, Prismatic-7B, MiniCPM-V-4.5-8B, Qwen3-VL, RoboRefer-8B, and RoboBrain2.5-8B. Although the strongest model, Qwen3-VL-32B, achieves 83.4% overall accuracy and 61.4 SBERT score, category-level results reveal persistent weaknesses in spatial reasoning, procedural reasoning, event prediction, and intention inference. ERQA-Plus therefore provides a fine-grained evaluation framework for measuring not only whether embodied agents answer correctly, but also which forms of embodied reasoning they can and cannot perform reliably. The dataset is available https://huggingface.co/datasets/huggingdas/erqa-plus and the project page at https://github.com/LUNAProject22/erqa-plus.

阅读与讨论 → 访问原文 →