探索全球前沿学术脉络

01.

arXiv (CS.AI) 2026-06-17 DOI: arXiv:2606.18168

All Smoke, No Alarm: Oracle Signals in Agent-Authored Test Code

作者:

Dipayan Banik↗Kowshik Chowdhury↗Shazibul Islam Shamim↗

arXiv:2606.18168v1 Announce Type: cross Abstract: Software practitioners increasingly use AI coding agents that generate test code alongside production code in open source pull requests (PRs). Recent studies report more than 932,000 agent-authored PRs across more than 116,000 repositories, yet whether their test files contain meaningful verification logic remains underexplored. Test files lacking explicit assertions execute code without verifying behavior, so quality gates based on test-file presence overestimate verification strength. The goal of this paper is to help practitioners assess the verification strength of agent-authored patches by characterizing oracle signals and their link to merge outcomes and review effort. We conduct an empirical study of 86,156 test-file patches from 33,596 agent-authored PRs across 2,807 GitHub repositories produced by five coding agents: OpenAI Codex, GitHub Copilot, Devin, Cursor, and Claude Code. A qualitative analysis of 384 stratified patches informs a syntactic taxonomy of eight oracle signal categories. Applied at scale, 80.2% of test patches contain weak or no explicit oracle signals. While raw merge rates are lower for strong-oracle PRs, a regression analysis adjusting for agent, PR size, repository popularity, task type, and language shows strong oracles significantly improve merge likelihood (OR = 1.28, p < 0.001). Our findings suggest that test file counts substantially overestimate verification strength and that practitioners can adopt oracle-aware quality checks to more accurately evaluate agent-authored contributions.

阅读与讨论 → 访问原文 →

02.

arXiv (CS.AI) 2026-06-19 DOI: arXiv:2511.08378

Bid Farewell to Seesaw: Towards Accurate Long-tail Session-based Recommendation via Dual Constraints of Hybrid Intents

作者:

Xiao Wang↗Ke Qin↗Dongyang Zhang↗Xiurui Xie↗Shuang Liang↗

arXiv:2511.08378v4 Announce Type: replace-cross Abstract: Session-based recommendation (SBR) aims to predict anonymous users' next interaction based on their interaction sessions. In the practical recommendation scenario, low-exposure items constitute the majority of interactions, creating a long-tail distribution that severely compromises recommendation diversity. Existing approaches attempt to address this issue by promoting tail items but incur accuracy degradation, exhibiting a "see-saw" effect between long-tail and accuracy performance. We attribute such conflict to session-irrelevant noise within the tail items, which existing long-tail approaches fail to identify and constrain effectively. To resolve this fundamental conflict, we propose HID (Hybrid Intent-based Dual Constraint Framework), a plug-and-play framework that transforms the conventional "see-saw" into "win-win" through introducing the hybrid intent-based dual constraints for both long-tail and accuracy. Two key innovations are incorporated in this framework: (i) Hybrid Intent Learning, where we reformulate the intent extraction strategies by employing attribute-aware spectral clustering to reconstruct the item-to-intent mapping. Furthermore, discrimination of session-irrelevant noise is achieved through the assignment of the target and noise intents to each session. (ii) Intent Constraint Loss, which incorporates two novel constraint paradigms regarding the diversity and accuracy to regulate the representation learning process of both items and sessions. These two objectives are unified into a single training loss through rigorous theoretical derivation. Extensive experiments across multiple SBR models and datasets demonstrate that HID can enhance both long-tail performance and recommendation accuracy, establishing new state-of-the-art performance in long-tail recommender systems.

阅读与讨论 → 访问原文 →

03.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2606.11931

Semantic Grading of Written Answers in Low-Resource Language Bangla Using a Fine-Tuned Lightweight Language Model

作者:

Meherun Farzana↗Aniket Joarder↗Mahmudul Hasan↗Md. Mosaddek Khan↗

Bangla is among the world's most widely spoken languages, yet it remains underserved in educational NLP research. In many remote and rural regions, access to qualified subject teachers is limited, and written answers are consequently graded largely by hand, restricting timely and consistent feedback. Automatic assessment is challenging because semantically correct responses can vary substantially in surface form. We present a bilingual (Bangla-English) evaluation system designed for low-resource educational settings that prioritizes semantic correctness over lexical overlap. Our approach fine-tunes a lightweight language model to grade each response using the question, reference answer, and student answer, producing a numeric score and concise, context-grounded feedback suitable for classroom deployment. We also construct a synthetic bilingual dataset to enable controlled training and evaluation. Across proprietary and open-source LLMs evaluated under a unified protocol, our QLoRA-tuned Qwen3-8B confirms consistent improvement by producing the most leakage-resistant feedback (RoRa = 0.819) in synthetic evaluation and the strongest agreement with human scores (rho = 0.936, MAE = 0.725) in a dedicated human study.

阅读与讨论 → 访问原文 →

04.

arXiv (CS.LG) 2026-06-19 DOI: arXiv:2505.18201

Reinforcement Twinning for Hybrid Control of Flapping-Wing Drones

作者:

Romain Poletti↗Lorenzo Schena↗Lilla Koloszar↗Joris Degroote↗Miguel Alfonso Mendez↗

arXiv:2505.18201v2 Announce Type: replace-cross Abstract: Controlling flapping-wing drones requires controllers that handle time-varying, nonlinear, underactuated dynamics from incomplete, noisy sensor data. Recent advances in artificial intelligence (AI), particularly reinforcement learning (RL), have opened new perspectives for addressing such complex control problems through data-driven policy optimization from interaction with the environment. Yet purely data-driven methods are sample-inefficient, demanding extensive, sometimes unsafe exploration, especially without guiding physical models. This motivates hybrid AI-physics frameworks. This article proposes a hybrid model-free/model-based flight-control approach using the reinforcement twinning algorithm. The model-based (MB) component uses an adjoint formulation and an adaptive digital twin continuously identified from live trajectories; the model-free (MF) component uses RL. The two agents share knowledge via transfer learning, imitation learning, and shared experience between the real environment and the digital twin, coordinated by a policy referee that selects which agent acts in reality based on digital-twin performance and a real-to-virtual consistency ratio. The framework is evaluated for the longitudinal control of a flapping-wing drone, modelled as a nonlinear time-varying system driven by quasi-steady aerodynamic forces. The hybrid strategy is tested under three adaptive-model initializations: (1) offline identification from existing data, (2) random initialization with fully online identification, and (3) offline pre-training with biased parameters followed by online adaptation. In all cases, the hybrid framework improves performance, robustness, and sample efficiency over purely model-free and purely model-based approaches.

阅读与讨论 → 访问原文 →

05.

arXiv (CS.LG) 2026-06-18 DOI: arXiv:2606.18430

Signature filtering: a lightweight enhancement for statistical watermark detection in large language models

作者:

Chih-Duo Hong↗Yen-Pang Chen↗Fang Yu↗

arXiv:2606.18430v1 Announce Type: new Abstract: Statistical watermarks help organizations attribute large language model (LLM) outputs, yet existing detectors often struggle when watermark signals are weak, texts are repetitive, or watermarks are edited. We propose signature filtering, a detection-time module that enhances watermark detection without modifying watermark embedding and text generation. It learns a small set of ``signature'' tokens whose presence makes watermark tests unreliable, and removes these tokens before detection. The signatures are obtained by solving a mixed-integer linear program on a small training set, with constraints that maximize the true positive rate. We additionally derive finite-sample and asymptotic bounds under several attacker models (color-blind, color-adaptive, and distributionally correlated). On four well-known watermark families (Kgw, Sweet, Unigram, Exp), four benchmark corpora (C4, MBPP, HumanEval, Code-Search-Net), and six LLMs (Opt-1.3b, Opt-6.7b, Llama2-13b, Llama3.1-8b, Qwen2.5-14b, Phi-3-medium-14b), 2- and 3-gram signatures raise detection rates in weak-signal and low-entropy settings from 8~31% without filtering to 78~99% with filtering, while keeping false positives controllable and often negligible. In stress tests where we scramble sentences and perturb 25~50% of tokens by dilution, deletions, and substitutions, 2-gram filters for Kgw-style watermarks preserve most of the clean-text detection gains, often matching or outperforming the advanced WinMax watermark detector. Signature filtering thus provides a simple, scalable, and model-agnostic add-on to strengthen watermark-based provenance checks for LLM text in information processing workflows.

阅读与讨论 → 访问原文 →

06.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2606.14943

Simplifying the Modeling of Arbitrary Conditionals in Natural Language

作者:

Yinhan Lu↗Eric Elmoznino↗L\'eo Gagnon↗Sarthak Mittal↗Tejas Kasetty↗Guillaume Lajoie↗

Causal Transformers model sequences through an autoregressive factorization of the joint distribution, which enables efficient left-to-right decoding and conditional likelihood computation. However, they cannot tractably sample from or evaluate arbitrary conditionals – e.g., a block of text conditioned on past and future tokens. Recent work aims to solve this problem through novel architectures, but they often lead to sub-optimal modeling of such conditionals and degraded generations. We propose Arbitrary Conditionals GPT (AC-GPT) which introduces a simple modification to standard causal Transformers to enable evaluating and sampling from arbitrary conditionals – including past, future, and mixed contexts – within a single forward pass. Unlike prior approaches, our method preserves the standard left-to-right ordering and next-token prediction objective essential for both strong performance and efficient training on natural language. Crucially, this compatibility allows existing LLMs to be fine-tuned for arbitrary conditioning. Our empirical results indicate that our method outperforms baselines on modeling arbitrary conditionals, without degrading standard left-to-right performance.

阅读与讨论 → 访问原文 →

07.

arXiv (math.PR) 2026-06-16 DOI: arXiv:2605.29641

Experimentation for Different Scheduling Policies on Queues: Mixed Differences-in-Q Estimators Based on Little's Law

作者:

Nanshan Jia↗Ramesh Johari↗Nian Si↗Zeyu Zheng↗

arXiv:2605.29641v2 Announce Type: replace-cross Abstract: In data centers, tasks are dispatched to various servers to evenly distribute the workload. When a data center considers implementing a new scheduling algorithm, it typically conducts an A/B test prior to deployment to assess the real-world impact of this new method. However, a straightforward A/B test might be interfered with so-called ``Markovian'' interference. We utilized the Differences-in-Q estimator, as developed by Farias et al. (2022), and introduced mixed Differences-in-Q estimators grounded in Little's Law. We show that our A/B testing methods significantly reduce bias and variance when testing various scheduling policies. Extensive simulations were conducted under scenarios like non-stationary arrival rates, heterogeneous service rates, and communication delays. These simulations highlight the robustness and efficacy of our A/B testing approach.

阅读与讨论 → 访问原文 →

08.

arXiv (CS.CL) 2026-06-17 DOI: arXiv:2606.17905

ChLogic: Evaluating Robustness of Logical Reasoning in Chinese Expressions

作者:

Peixian Zhou↗Yuxu Chen↗Chaorui Zhang↗Wei Han↗Bo Bai↗Xueyan Niu↗

Large language models perform increasingly well on standardized logical reasoning benchmarks, but whether this ability remains robust beyond English is unclear. We introduce ChLogic, an English–Chinese aligned benchmark that tests whether models preserve logical reasoning performance when the same latent logical structure is expressed in English and diverse Chinese surface realizations. Built from formal logical templates, the benchmark contains three data sets: (i) the General aligned set, derived from 60 General Propositions across nine template families; (ii) the Difficult aligned set, derived from 40 Difficult Problems; and (iii) the Chinese-only set, covering 15 language-specific phenomenon types. Each aligned item pairs one English reference expression with five Chinese realizations. Experiments on Qwen3, Ministral, and GLM models reveal a persistent English–Chinese performance gap. Back-translation from standard Chinese into English often improves performance on the General aligned set, but produces mixed effects on the Difficult aligned set, where Qwen3-32B and GLM-5.1 perform worse after translation. These results indicate that Chinese surface realization, translation artifacts, and model-specific behavior jointly affect multilingual logical reasoning. Overall, ChLogic provides a useful stress test for the robustness of multilingual reasoning.

阅读与讨论 → 访问原文 →

09.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2606.12069

Tac-DINO: Learning Vision-Tactile Features with Patch Alignment

作者:

Hong Li↗Yankang Dong↗Yue Xu↗Yihan Tang↗Mingzhu Li↗Jiamin Qiu↗Qihang Yao↗Xing Zhu↗Yujun Shen↗Nan Xue↗Yong-Lu Li↗

Touch is the primary medium through which humans interact with the environment. Currently, tactile learning mainly focuses on image-level pretraining or alignment. However, tactile signals correspond to local object contact, while research into scale alignment and holographic matching remains limited and proper datasets and benchmarks also lack. To bridge this gap, we first construct a data collection system to acquire a large-scale tactile dataset, with over 20 K tactile contacts from 505 real-world objects. Building on this dataset, we design a Vis-Tac Holographic Matching Benchmark to evaluate vision-tactile local-to-global alignment ability. Then we propose Vision-Tactile Patch Alignment (VTPA) methods for vision-tactile representation learning. Experiments demonstrate that these exceed the performance of methods without alignment and align with whole-object images.

阅读与讨论 → 访问原文 →

10.

arXiv (CS.LG) 2026-06-11 DOI: arXiv:2606.11743

TacCoRL: Integrating Tactile Feedback into VLA via Simulation

作者:

Siyu Ma↗Yuqi Liang↗Chang Yu↗Yunuo Chen↗Hao Su↗Yixin Zhu↗Yin Yang↗Chenfanfu Jiang↗

arXiv:2606.11743v1 Announce Type: cross Abstract: Vision-language-action (VLA) models provide strong visual, language, and action priors for robot manipulation, but visual observations alone often miss the local contact state required for contact-rich tasks. We present TacCoRL, a scalable framework that injects Tactile feedback into VLA policies and improves them through sim-real Co-training and simulation-based reinforcement learning (RL), without requiring large-scale tactile pretraining or extensive real-world contact exploration. The key idea is not only adding touch as an input, but learning how contact readings should modulate action responses in near-failure states that are rare in demonstrations and risky to collect on hardware. We use a real-aligned simulator as a closed-loop training environment for contact interaction. Mixed simulated and real trajectories first warm-start tactile-conditioned actions in the pretrained policy. Reinforcement learning with verifiable task rewards then optimizes the policy using simulated contact rollouts. It reinforces tactile-conditioned actions that lead to task completion, while a supervised objective on real trajectories keeps the refined policy anchored to deployment visual, tactile, and action distributions. The resulting policy transfers directly to the real robot without privileged simulation state or online real-world RL. Across four bimanual contact-rich tasks, the final visuo-tactile policy achieves an average success rate of 72.5%, compared to baseline of 50.0%. Result videos and more details are available at https://tac-corl.github.io/

阅读与讨论 → 访问原文 →

11.

arXiv (CS.LG) 2026-06-11 DOI: arXiv:2606.11258

Loss Landscape Diagnosis for Gradient-Based Gray-Scott System Inversion: Disentangling the Roles of PINN Components

作者:

Yan Yang↗

arXiv:2606.11258v1 Announce Type: new Abstract: Gradient-based inversion of reaction-diffusion systems is typically approached via surrogate models or physics-informed neural networks (PINNs), while the most direct route, backpropagation through the PDE's structure itself, has largely been avoided. We pursue this direct route as a diagnostic probe, backpropagating a steady-state loss through unrolled Gray-Scott simulation to recover its parameters, with no surrogate or neural-network augmentation. Optimization fails to converge, and plotting the landscape directly locates the failure in its geometry – flat plateaus with no gradient signal, bounded by sharp cliffs that align with bifurcation boundaries – a structure that recurs across loss functions and is inherited however the gradients are routed to parameters. Reading this minimal setup as an ablation of PINN, we disentangle each component's role: with the neural network fixed, the residual loss is quadratic in the PDE parameters and yields a smooth landscape, so it alone already avoids the pathology, by implicitly encoding the full PDE dynamics across all initial conditions. The neural network, for its part, cannot repair an ill-posed parameter subspace, and so serves only to complete the observed data – a division of labor not previously made explicit. These findings carry concrete design implications for PINN-type methods and a broader heuristic on when added dimensions actually help.

阅读与讨论 → 访问原文 →

12.

medRxiv (Medicine) 2026-06-22 DOI: HASH:d7d811ef71b29bb4186c3590d8acc97a

Cumulative Metabolic Exposure to Hyperglycemia and Risk of Cardiovascular and Limb Events in Peripheral Artery Disease

作者:

James↗Li↗Somisetty↗Nguyen↗Vaughan-Sarrazin↗M. S↗Lund↗Tsai↗Smolderen↗K. G↗Hoffman↗R. M↗…

Background: Although diabetes is a potent risk factor for the development of peripheral artery disease (PAD), the effect of cumulative metabolic exposure to hyperglycemia on risk of cardiovascular or limb events in patients with PAD remains unclear. Methods: The Peripheral Artery Disease: Long-term Survival (PEARLS) is a longitudinal registry of Veterans with newly diagnosed PAD identified using a natural language processing approach. Included patients had ankle brachial index [&le;]0.9 or toe brachial index [&le;]0.7, and no history of lower extremity revascularization or major amputation. Among patients with diabetes in this cohort, we assessed cumulative exposure to hyperglycema based on a 24-month rolling average of hemoglobin (Hgb) A1c values, categorized as [&le;]7%, >7% to [&le;]8%, and >8%. Multivariable Cox regression models evaluated the association between categories of HgbA1c, modeled as a time-varying exposure, and risk of cardiovascular (CV: myocardial infarction or stroke) and limb (chronic limb threatening ischemia [CLTI] or major amputation) events. Results: Among 45,109 patients with new diagnosis of PAD and pre-existing diabetes, the mean HgbA1c at baseline was 7.5%, with nearly one-third (30.4%) having HgbA1c >8%. The mean age was 70.4 years, 19.8% were Black and 4% were Hispanic. Patients with baseline HgbA1c >8% were younger and compared to those with HgbA1c [&le;]7%, more likely to have coronary disease, kidney disease, and obesity. Over a median follow up of 4.2 years, 8,306 (18.4%) patients experienced a CV event, and 8,199 (18.2%) experienced a limb event. The adjusted association between HgbA1c and hazard of CV events was 12% higher in patients exposed to HgbA1c >7% to [&le;]8% (HR 1.12; 95%CI: 1.05-1.18) and 38% higher in those exposed to HgbA1c >8% (HR 1.38; 95%CI: 1.30-1.46), compared to HgbA1c 7% to [&le;]8% (HR 1.20; 95%CI: 1.13-1.28) and HgbA1c >8% (HR 1.60; 95%CI: 1.51-1.70), respectively when compared to HgbA1c [&le;]7%. These findings were consistent in subgroups based on age and severity of PAD. Conclusions: Among diabetic patients with PAD, cumulatiave metabolic exposure to hyperglycemia is associated with a markedly increased risk of clinical events, especially limb events.

阅读与讨论 → 访问原文 →

13.

bioRxiv (Bioinfo) 2026-06-11 DOI: HASH:7d877a310c7064fcfc6915949d100b91

DeePEn - A Depth sensitive benchmark for Protein Engineering

作者:

Schmirler↗Brenner↗Franz↗Heinzinger↗Rost↗

Recent progress in modeling techniques and high-throughput screening has significantly enhanced the accessibility of protein engineering. Nevertheless, further progress gets hindered by the lack of robust benchmarks that capture the practical challenges for real-world protein engineering. Here, we introduced DeePEn, a Depth-sensitive benchmark for Protein Engineering that quantifies a models generalization capabilities when predicting protein fitness at increasing mutational distance from the wildtype or training data. We defined distance as the number of simultaneous point mutations, i.e., single amino acid variants (SAVs), moving from wild-type to mutant (edit distance in computer science jargon). Specifically selecting four deep mutational scanning (DMS) datasets with sufficient multi-mutation data points from ProteinGym, we assessed recent predictive models, including general and biophysics-informed protein Language Models (pLMs), and a non-transformer neural network. Our results highlight how the performance of all models deteriorates with increasing mutational distance and that no single metric sufficiently captures the diverse requirements of protein engineering. To overcome these shortcomings, DeePEn provides a readily available resource for multi-metric benchmarking that focuses on the prediction of distant variants.

阅读与讨论 → 访问原文 →

14.

arXiv (CS.LG) 2026-06-19 DOI: arXiv:2606.19770

An Information Theoretic Framework for Graph Novelty Generation via Latent Mixture Modeling

作者:

Itsuki Nakagawa↗Kenji Yamanishi↗

arXiv:2606.19770v1 Announce Type: new Abstract: We propose an information-theoretic framework for graph novelty generation, which aims to generate data that are distinct from existing patterns while preserving global structural consistency. Our approach embeds data into a latent space, models the latent distribution using finite mixture models, and generates novel samples by imposing explicit novelty and reliability conditions formulated in terms of description length. Specifically, novelty is enforced by requiring generated samples to be poorly explained by all existing mixture components, while reliability constrains their impact on the overall mixture structure under the Minimum Description Length (MDL) principle. We provide a theoretical analysis showing that, with appropriate threshold choices, the probabilities of misclassifying non-novel or unreliable samples converge to zero with explicit rates. Experiments on synthetic and benchmark graph datasets demonstrate that the proposed method enables principled novelty generation with quantifiable risk.

阅读与讨论 → 访问原文 →

15.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2606.12263

VOID: Defeating Unauthorized Mimicry in Latent Diffusion Models

作者:

Chunlin Qiu↗Ang Li↗Tianxiao Huang↗Ruilin Gan↗Yunjie Ge↗Shenyi Zhang↗Huayi Duan↗Lingchen Zhao↗Chao Shen↗Qian Wang↗

While Latent Diffusion Models (LDMs) have revolutionized visual synthesis, they are increasingly exploited for unauthorized mimicry of individuals. Existing defenses inject deceptive perturbations to steer the generated images toward irrelevant targets. However, this approach hinges on an ungrounded assumption: subtle perturbations can maintain their deceptive efficacy throughout an LDM's extensive generation process. In reality, the model's innate restoration mechanism will remove such perturbations and cause individual identities to re-emerge in the images generated. We propose VOID, a defense framework that overcomes this conundrum by manipulating an LDM's intrinsic stochasticity. VOID perturbs the diffusion pipeline in two novel ways: 1) amplifying the latent encoding errors to shatter an image's semantic structure, and 2) counteracting the target guidance signals to suppress the model's restoration capabilities. This results in a semantic corruption that thwarts any unauthorized mimicry. Notably, the security gain does not come at the price of visual utility, as VOID simultaneously manages to confine perturbations to human-imperceptible regions of protected images. Our comprehensive evaluation of 24 state-of-the-art defenses against 10 mimicry attacks on 5 datasets demonstrates VOID's unprecedented protection power: it increases the average Frechet Inception Distance (FID) from 113 to 365, a 223% improvement over the strongest defense to date.

阅读与讨论 → 访问原文 →

16.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2506.20040

Cross-Layer Discrete Concept Discovery for Interpreting Language Models

作者:

Ankur Garg↗Xuemin Yu↗Hassan Sajjad↗Samira Ebrahimi Kahou↗

Interpreting language models remains challenging due to the existence of residual stream, which linearly mixes and duplicates features across adjacent layers, causing single-layer analyses to miss this cross-layer structure. Cross-layer sparse autoencoders (SAEs) address layer mixing but operate in continuous space, where concepts split across many neurons without clear boundaries. We introduce Cross-Layer Vector Quantized-Variational Autoencoder (CLVQ-VAE), a novel framework which maps representations from a lower layer to a higher layer through a discrete vector-quantization bottleneck, collapsing duplicated residual-stream features into compact, interpretable concept vectors. Our approach combines top-k temperature-based sampling with exponential moving average (EMA) codebook updates, providing controlled exploration of the discrete latent space while maintaining codebook diversity. Across both encoder- and decoder-based models on ERASER-Movie, Jigsaw, and AGNews, CLVQ-VAE outperforms clustering, single-layer vector quantized-variational autoencoder (VQ-VAE), and sparse autoencoder (SAE) baselines across three evaluation axes: removing identified concepts drops model accuracy by up to 93%, LLM judges rank our concepts first in 66.7% of comparisons, and human annotators recover model predictions from our visualizations with 78% accuracy versus 54% for clustering.

阅读与讨论 → 访问原文 →

17.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2605.17106

HyDRA: Hybrid Dynamic Routing Architecture for Heterogeneous LLM Pools

作者:

Aashna Garg↗Siddharth Singha Roy↗Jinu Jang↗Federico Brancasi↗Shengyu Fu↗

Production LLM deployments increasingly maintain heterogeneous model pools spanning order-of-magnitude cost differences. Existing routers make binary strong-vs-weak decisions and couple learned parameters to specific model identities, requiring retraining whenever the catalog changes. We present HyDRA (Hybrid Dynamic Routing Architecture), a framework that predicts fine-grained, multi-dimensional capability requirements per query and matches them against configuration-defined model profiles via shortfall matching. A ModernBERT encoder with K=4 independent sigmoid heads scores each query along reasoning, code generation, debugging, and tool use; a shortfall-matching algorithm then selects the cheapest model whose capabilities meet the predicted requirements. The deployed predictor runs at 86 ms median CPU inference latency in production, and is fully decoupled from the model catalog – adding or removing models requires only a configuration change, with zero retraining. On SWE-Bench Verified (5-model pool: GPT-5.4-mini, Claude Haiku 4.5, GPT-5.3 Codex, Claude Sonnet 4.6, GPT-5.4), HyDRA's tunable shortfall threshold spans three regimes: peak-quality exceeds the always-strong Claude Sonnet 4.6 baseline (75.4% vs. 74.2% resolution) at 12.9% cost savings; iso-quality matches Sonnet at 54.1% cost savings, a 6x improvement over our prior in-house binary router at 9.1%; aggressive pushes savings to 72.5% for a 3.2-point quality trade. Results generalize across LiveCodeBench, BigCodeBench, and tau-bench. HyDRA is deployed to all users in GitHub Copilot's VS Code Chat auto-mode and – to our knowledge for the first time in the LLM routing literature – demonstrates language-invariant routing across CJK, European, and other script families.

阅读与讨论 → 访问原文 →

18.

bioRxiv (Bioinfo) 2026-06-18 DOI: HASH:04438477b0d11f56ec944226861ae7d5

A Two-Stage Interpretable Framework for Predicting Plant-Derived Small RNA Targets on Human 3'UTRs

作者:

qiao↗li↗

Can plant-derived small RNAs target human mRNA 3'UTRs via complementary base pairing and produce experimentally detectable regulatory effects? This question concerns not only the fundamental feasibility of cross-kingdom RNA regulation but also the technological pathway for screening plant-derived active small nucleic acids. Existing miRNA target prediction tools are predominantly designed for endogenous miRNA-mRNA systems, exhibiting notable limitations when applied to cross-species small RNA inputs and small-sample wet-lab experimental adaptation. In this study, we developed a two-layer prediction framework, MetaLulu-AI. The first layer builds upon publicly available human miRNA-mRNA 3'UTR interaction data, utilizing XGBoost to learn foundational binding rules on human 3'UTRs based on 41 interpretable computational features, including seed region pairing types, local context sequence composition, site positioning, and RNA secondary structures. The second layer is tailored to the experimental system of plant-derived small RNAs and human target genes. It introduces 40 experimental samples using significant changes in endogenous protein expression as the regulatory standard (determined by Western blot or ELISA 48 hours post-transfection of small RNAs via Lipo3000). Using 52-dimensional computational features and the optimal transcript scores from the first layer as inputs, this layer employs TabPFN for experimental label adaptation. The first-layer dataset consists of 38,752 training samples, 5,536 validation samples, and 11,073 testing samples (totaling 55,361), with a positive-to-negative sample ratio of approximately 1:5.4. On the randomly split test set, the model achieved an AUC of 0.9686, a recall of 0.8523, a precision of 0.8080, and an accuracy of 0.9452 (at a decision threshold of 0.4797). Group-based splitting revealed that the model maintains high discriminative power for unseen genes (AUC = 0.9541), though its generalization ability for completely unseen miRNAs decreases (AUC = 0.7390). For the 40 experimental samples in the second layer, the TabPFN model achieved an average AUC of 0.7406 {+/-} 0.092 across ten repeated 70/30 random splits, outperforming the baseline of directly using the first-layer scores (0.3563 {+/-} 0.149); the average AUC in a 5-fold cross-validation was 0.770 {+/-} 0.177. SHAP analysis demonstrated a clear divergence in the discriminative basis of the two models: the first layer relies more heavily on the thermodynamics of the small RNA itself and the quality of canonical seed sites, whereas the second layer focuses more on the local UTR environment and statistical site features. Although the current second-layer results are constrained by sample size and gene coverage, this framework serves as a preliminary observation of the adaptation mechanism for cross-kingdom regulation experiments, and motivating future large-scale validation. Under stricter leave-one-gene-out and leave-one-small-RNA-out evaluation, the adapter exceeded the first-layer score baseline but only matched the majority-class baseline, underscoring that entity-level generalization is not yet established.

阅读与讨论 → 访问原文 →

19.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2606.11337

Can AI Agents Synthesize Scientific Conclusions?

作者:

Hayoung Jung↗Pedro Viana Diniz↗Jos\'e Reinaldo Corr\^ea Roveda↗Abner Fernandes da Silva↗Haeun Jung↗Enoch Tsai↗Aleksandra Korolova↗Manoel Horta Ribeiro↗

Scientific AI agents increasingly retrieve evidence, reason across sources, and synthesize conclusions used in consequential decisions. Yet, their ability to do so in high-stakes domains such as health remains unclear. We introduce SciConBench, a large-scale live benchmark of 9.11K questions and expert-written conclusions from systematic reviews to evaluate open-domain scientific conclusion synthesis. The benchmark draws on an expert-validated automated evaluation pipeline that decomposes conclusions into atomic facts and measures correctness and comprehensiveness via factual precision and recall. To mitigate data leakage, we further introduce SciConHarness, a clean-room evaluation harness that equips agents with controlled web interaction to ensure valid measurement. Evaluating 8 frontier models and deep research agents, we find that factual quality remains low: under clean-room settings, the best agent achieves only a factual F1 of 0.337. Our clean-room setting consistently reduces performance relative to unconstrained evaluation, suggesting that leakage inflates estimates of models' true synthesis capabilities. Finally, we audit consumer-facing agents (e.g., Google AI Overview, OpenEvidence) and find they frequently generate incomplete and sometimes contradictory conclusions, even when the ground-truth answer is available. Overall, our results show that reliable synthesis of scientific conclusions remains an open challenge, and that clean-room evaluation is essential for assessing open-domain AI agents.

阅读与讨论 → 访问原文 →

20.

arXiv (CS.CL) 2026-06-15 DOI: arXiv:2606.13686

Benchmarking Web Agent Safety under E-commerce Deceptive Interfaces

作者:

Zijing Shi↗Meng Fang↗Ling Chen↗

As autonomous web agents are increasingly deployed to perform real-world tasks, ensuring their safety has become a critical concern. In this work, we study web agent behavior under realistic deceptive interfaces in the e-commerce domain. We introduce WebDecept, a lightweight and configurable plugin framework that enables controlled injection of deceptive interface patterns into existing web environments. Using WebDecept, we instantiate seven deceptive patterns commonly observed on the open web, including targeted advertisements, domain redirection, and shopping manipulation. By injecting these patterns into the frontend during task execution, we perform controlled evaluation of multiple multimodal web agents. Our results show that current web agents are highly susceptible to multiple classes of deceptive interfaces, and that prompt-based constraints are often insufficient to mitigate these failures. We further analyze how the design choices of deceptive patterns influence the success of such manipulations. These findings highlight safety challenges that should be addressed as web agents are scaled toward real-world deployment.

阅读与讨论 → 访问原文 →

21.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2606.16073

Stop the Sampler! Classifier-Based Adaptive Stopping for Sampling Kernels

作者:

Kirill Korolev↗Nikita Morozov↗Stepan Pavlenko↗Esmeralda S. Whitammer↗Sergey Samsonov↗

arXiv:2606.16073v1 Announce Type: new Abstract: Sampling from complex, unnormalized probability densities is a fundamental challenge in Bayesian inference and probabilistic modeling. While Markov chain Monte Carlo (MCMC) methods provide asymptotic guarantees, they often suffer from slow mixing and high computational costs due to fixed or manually tuned trajectory lengths. In this work, we propose a novel framework that treats trajectory termination as a learnable component of the sampling dynamics. By framing MCMC within the theory of non-acyclic generative flow networks (GFlowNets), we train state-dependent neural classifiers to decide when a trajectory has reached a high-density region and should terminate. We theoretically establish the connection between optimal classifiers and the target density via detailed balance conditions and introduce a multilevel training scheme to facilitate exploration in complex geometries. Experimental results across various benchmark densities demonstrate that our approach significantly reduces average trajectory lengths while improving mode coverage and mixing compared to standard MCMC baselines.

阅读与讨论 → 访问原文 →

22.

arXiv (CS.CL) 2026-06-17 DOI: arXiv:2606.09376

Precision Is Not Faithfulness: Coverage-Aware Evaluation of Grounded Generation with a Complete Oracle

作者:

Juan S. Santillana↗

Reference-free faithfulness metrics verify each atomic claim a model makes against ground truth, and are increasingly used to evaluate grounded generation. We show they share a blind spot: they measure only precision – are the stated claims supported? – and therefore reward abstention, since a model can score near-perfect faithfulness by saying almost nothing. We make this measurable using Formula 1 telemetry, a domain where strategic ground truth is derived deterministically and, crucially, completely: for each decision we know the full set of facts that mattered. This completeness – absent in open-domain faithfulness benchmarks – lets us measure recall (coverage of the relevant facts) exactly, alongside precision. On a multilingual (EN/ES/PT) benchmark of 7,253 decision instances spanning 157 races, the most precise frontier model covers under half of the relevant facts and ranks last by F1, so requiring coverage reorders the systems; the same effect reappears in a second complete-oracle domain (NOAA weather forecasts). Fine-tuning small models (1B-7B) on the complete oracle closes the precision-recall gap entirely (F1 ~0.98), beating every zero-shot frontier system regardless of scale. We pair faithfulness with coverage into a single score, validate the metric (controlled perturbation; agreement across a model-free regex extractor and a cross-family LLM extractor, system-level Spearman 1.0), and give a verifier-guided generation method that improves precision and recall without references. We release the benchmark, structured annotations, metric, baselines, and an interactive demo.

阅读与讨论 → 访问原文 →

23.

arXiv (quant-ph) 2026-06-15 DOI: arXiv:2606.14624

Spin disorder competing with positional symmetry breaking governs the metal-insulator behavior in oxide paramagnets

作者:

Jia-Xin Xiong↗Xiuwen Zhang↗Alex Zunger↗

arXiv:2606.14624v1 Announce Type: cross Abstract: Numerous transition-metal oxides have low-temperature antiferromagnetic (AFM) states and high-temperature paramagnetic (PM) phases, where the AFM state is usually insulating while the PM phase can be either insulating or metallic. Without involving strong correlation, we use symmetry-broken density-functional theory (DFT) to obtain the PM phases of insulating NaFeO3 vs the recently discovered metallic NaOsO3. We develop the understanding of insulating and metallic behaviors in paramagnetic oxides by analyzing the interactions between magnetic and positional symmetry breaking: The insulating gap is governed by the competition between the spin disorder that induces a distribution of different magnitudes of local magnetic moments and the polymorphous distribution of off-center atomic displacements. NaFeO3, on the other hand, has large positional displacement with small spin-disorder-induced moments distribution, leading to insulating PM phase, whereas NaOsO3 has a pronounced spin-disorder-induced moments distribution that forces the PM phase to become metallic. Our work identifies this symmetry-breaking competition as a general framework to bridge seemingly disparate metal-insulator behaviors in transition-metal oxides paramagnets without invoking strong correlation.

阅读与讨论 → 访问原文 →

24.

arXiv (CS.CV) 2026-06-19 DOI: arXiv:2606.19889

SurgVista: Long-Horizon Surgical World Modeling with Plausible Instrument-Tissue Dynamics

作者:

Wentao Pan↗Wuyang Li↗Shengyuan Liu↗Xinyu Liu↗Hengyu Liu↗Yixuan Yuan↗

Scaling robot policy learning for autonomous surgery is challenging, as expert demonstrations are expensive and in vivo exploration poses substantial safety risks. Surgical world models address this by generating realistic, action-conditioned future frames from an initial observation, but existing methods exhibit two persistent failure modes: spatial interaction incoherence, where visible instrument contact fails to induce spatially consistent tissue deformation, and temporal fidelity collapse, where prediction errors compound across autoregressive rollouts and progressively corrupt visual quality. We present SurgVista, a surgical world model that mitigates both failures through two training recipes. Deformation Consistency Regularization extracts scene-point trajectories from training videos and enforces cross-frame coherence through latent contrastive learning, strengthening physically consistent instrument-tissue dynamics. Drift Adaptation Training mitigates long-horizon drift by perturbing conditioning frames with online prediction residuals and photometric augmentations calibrated to long-horizon drift statistics, sustaining visual fidelity over extended rollouts. To enable rigorous evaluation, we further introduce SurgWorld-Bench, featuring diverse procedure types, long-range rollouts, and decoupled metrics for instrument-motion accuracy and tissue-response fidelity. Extensive experiments show that SurgVista consistently outperforms state-of-the-art methods across visual quality, temporal consistency, and interaction fidelity, with gains widening as the prediction horizon grows.

阅读与讨论 → 访问原文 →

25.

arXiv (CS.CL) 2026-06-17 DOI: arXiv:2604.13899

Do We Still Need Humans in the Loop? Comparing Human and LLM Annotation in Active Learning for Hostility Detection

作者:

Ahmad Dawar Hakimi↗Lea Hirlimann↗Isabelle Augenstein↗Hinrich Sch\"utze↗

Instruction-tuned LLMs can annotate thousands of instances at low cost. This raises two questions for active learning (AL): can LLM labels replace human labels within the AL loop, and does AL remain necessary when entire corpora can be cheaply labeled? We investigate both on a new dataset of 277,902 German political TikTok comments (25,974 LLM-labeled, 5,000 human-annotated), comparing LLM and human annotation across seven conditions, four encoders, and 10 random seeds. Under a two-question interface that mirrors the human annotation task, LLM annotation at scale outperforms human-supervised classifiers at roughly one-tenth the cost (\$28 for GPT-5.2 Batch API vs. \$316 for Prolific). The advantage holds for both a closed-source (GPT-5.2) and an open-weight (Qwen3.5-122B-10B) LLM, is robust under soft-label evaluation, and is unlocked specifically by the two-question decomposition; a holistic single-prompt baseline only ties with human supervision. AL provides no reliable advantage over random sampling under either LLM annotator. However, error structure varies sharply: only GPT-5.2 under the two-question interface produces classifiers with near-human FP/FN balance, while other LLM variants over-flag border-control and economic competition discourse. We release the dataset and code.

阅读与讨论 → 访问原文 →