Named after the hundred-eyed watchman of Greek myth, Argus watches the education landscape: spotting new opportunities, pressure-testing the ventures we're building, and tracing every read back to the real-world signals behind it.
The evidence library: the raw signals the pipeline is watching across the education ecosystem. Every idea is built from these.
arXiv:2607.12755v1 Announce Type: cross Abstract: AI-enabled systems are seeing increasing deployment across numerous domains, with many being "black boxes" with respect to core functions and capabilities. I.e., many systems take inputs and give outputs, but without users having any ability to see how the former lead to the latter. AI-enabled systems are also being used to augment autonomy in systems, and autonomy coupled with opacity raises numerous concerns surrounding, e.g., the reliability of systems, their regularity in functioning, human ability to control them, or whether deploying opaque and potentially autonomous systems is in compliance with ethical and legal norms. In this article, we argue that many of these worries can be mitigated by leveraging practical judgment, virtue, and intuition in the deployment and use of opaque AI-enabled systems. We show that focusing on these distinctly human capabilities provides a means for bridging between the practical challenges created b
arXiv:2607.12650v1 Announce Type: cross Abstract: Tool access alone does not make LLM empirical reasoning governable: accepted outputs need not descend from attested evidence, and accepted deductions need not hold up under formal scrutiny. We present EG-VAR (Evidence-Grounded Verified Agentic Reasoning), a Lean 4-based tool-calling architecture in which the Lean kernel is the sole minter of Verified claims via tool-attestation axioms and declared source lifts. Every verified output structurally descends from an attested tool call (Thm. 3.1) and a kernel-checked chain of valid inference (Thm. 3.2); residual outputs are honest Abstain with a replayable audit trail. On a subcollection of TableBench numerical reasoning (n=120), EG-VAR attains 120/120 versus a 95% same-tool baseline; on counterfactual stress tests (5 domains x 2 models), EG-VAR stays 100% source-faithful while same-tool drops to 80-90% (no-tool 50-80%). With the LLM as deployment-time formalizer, residual semantic-formaliza
arXiv:2607.12575v1 Announce Type: cross Abstract: AI agents are said to be forming an economy in which they pay, on their own, for the data, APIs, and compute they consume. x402, which settles a stablecoin payment on-chain for each purchase, is the most widely deployed protocol for this, and its hundreds of millions of settlements are read as proof that the economy has arrived. We show the count cannot be read as adoption: it is the one metric an interested party can manufacture almost for free, since the facilitator sponsors the gas and nothing on-chain marks who controls a payment. We give the first population-scale measurement of x402 on Base, supplemented with a coarser Solana census. Identifying settlements from their on-chain event and resolving the true payer through the meta-transaction layer, we sort each by what its trace can prove via a payment graph. Over a 280-day window Base carries 136{,}708{,}672 settlements worth \$44{,}121{,}383.81, concentrated on every axis we measu
arXiv:2607.12336v1 Announce Type: cross Abstract: Artificial Intelligence (AI) technologies, while serving as a foundational enabler for modern social media and digital health services, exert a bivalent effect by simultaneously acting as a combatant against and a spread vector for misinformation. A prevalent challenge in mitigating this issue arises in non-English contexts and low socioeconomic classes, where limited data hinders the training of AI models for effective detection. Consequently, culturally and linguistically diverse (CALD) communities struggle to access trustworthy health information through AI-driven tools. Current AI tools underperform due to a lack of training data and are largely unable to consider language nuances and traditions in non-English contexts. This research addresses these gaps by proposing a CALD-friendly AI-based health misinformation detector and providing a dashboard for medical professionals to analyse this misinformation, a critical step toward mitig
arXiv:2607.12298v1 Announce Type: cross Abstract: Convolutional neural networks (CNNs) are increasingly being deployed on system-on-chip (SoC) platforms, where hardware-accelerated inference enables low-latency edge computing. Achieving fault tolerance on these devices remains challenging because conventional redundancy (dual/triple modular redundancy, DMR/TMR) incurs high resource cost, while software-centric methods (e.g., algorithm-based fault tolerance (ABFT), checkpoint-restart, instruction-level duplication, and software watchdogs/assertions) introduce nontrivial latency/energy overheads, reduce model accuracy, or provide inadequate coverage for accelerator-induced faults. In this paper, we propose Emulated Integrity Replica (EIR), a hierarchical digital-twin framework for FPGA SoCs that provides autonomous fault detection and recovery. Unlike DMR/TMR, which replicates hardware logic and incurs proportional area and power overheads, EIR avoids fabric-level duplication by exploiti
arXiv:2607.12200v1 Announce Type: cross Abstract: As frontier language models advance, policymakers and model developers need methods for assessing whether model access materially increases a non-expert actor's ability to plan high-consequence Chemical, Biological, Radiological, or Nuclear (CBRN) misuse relative to public tools alone. Existing CBRN evaluations differ in non-expert definitions, threat scope, baselines, scoring rubrics, and decision rules, making results difficult to compare across studies. We introduce a Threshold Exceedance Criteria (TEC) framework that decomposes an uplift study into independently executable components: determining non-expert participant eligibility, defining the CBRN threat scope for the study, and statistically estimating material uplift. We then operationalize the TEC framework in a large-scale empirical study using a design that determines two forms of uplift: generative (where a model assists plan creation from scratch) and revisionist (where a m
arXiv:2607.12086v1 Announce Type: cross Abstract: Recent LLM-based multi-agent urban simulators can generate semantically rich city routines, but they remain costly to scale and are often weakly validated against empirical mobility patterns. We present CityBehavEx, an interactive LLM-assisted urban simulation platform that scales to city-size populations, exposes agent behavior for inspection, supports empirical validation, and generates mobility patterns that better match real-world spatial, temporal, and semantic distributions. Instead of invoking large language models for every agent action, CityBehavEx combines established human mobility models with fine-tuned cross-encoders that estimate semantic alignment between agent profiles, schedules, and activity transitions. This design enables large-scale simulations, as demonstrated in a case study of 100,000 agents over 75 days in under one hour on a single consumer GPU. The platform allows users to define simulation regions, launch exp
arXiv:2607.11918v1 Announce Type: cross Abstract: Dual submissions, in which identical or substantially similar papers are simultaneously submitted to one or more archival venues, without cross-citation or disclosure, are a growing problem for the AAAI Conference and other scientific publication venues. These submissions increase the burden on the peer-review system and pollute the scientific record. As part of the AAAI-26 review process, we (conference organizers) compared AAAI main-track submissions to nine other archival venues with overlapping review periods. We also searched for dual submissions within the AAAI-26 main track. We employed title+abstract similarity assessment to prioritize highly similar paper pairs for subsequent triage by an LLM-based overlap assessment tool, followed by manual review of the highest severity pairs. Manual review of such pairs led to the desk-rejection of 141 AAAI-26 main-track submissions. We seek to alert future organizers, and the broader artifi
arXiv:2607.12296v1 Announce Type: new Abstract: With the increased use of generative AI (GenAI) applications such as ChatGPT, higher education institutions (HEIs) have released a range of guidelines and policies to direct adoption within their institutions. In computer science (CS) courses GenAI adoption is especially high and the implications for student learning are significant. At the same time, instructors have also been forced to address the use of GenAI as students have started to use it for a range of functions. Currently, comparative analysis of guidance provided by institutions and its uptake in instruction is lacking. In this paper we bridge this gap by comparing institutional and computing course level guidance to better understand this terrain. We utilize secondary analysis of institutional and course syllabi guidelines from higher education institutions in the U.S. classified as research-intensive. Our findings reveal that although institutional guidance is more pro-use, a
arXiv:2607.12295v1 Announce Type: new Abstract: The rapid integration of artificial intelligence (AI) and generative AI (GenAI) into education presents significant opportunities to enhance teaching and learning, while raising ethical concerns about the responsible use of these technologies in educational settings. Understanding how the public perceives and debates these issues is increasingly important for educators, institutions, and policymakers seeking to integrate AI responsibly and equitably. Social media platforms, where such debates unfold frequently and at scale, offer a valuable lens for capturing large-scale, real-time public reactions to key developments as they emerge. In this study, we analyse five years (2019-2024) of discourse on Twitter (now X) to trace the evolving public conversation around AI ethics in education, paying particular attention to the release of ChatGPT as a pivotal moment that reshaped the nature and tone of that discourse. Using BERT-based topic modell
arXiv:2607.12235v1 Announce Type: new Abstract: This study proposes a semi-automated system for generating dialogue-based lessons using Large Language Models (LLMs) and Text-to-Speech (TTS) technology, and exploratorily examines its educational potential via a practical quasi-experiment. The system augments rather than replaces educators through a three-stage human-in-the-loop workflow (LLM-based slide/narration generation, educator review, automated audiovisual integration), and introduces a novel method for generating Expert-Novice dialogue narration based on cognitive apprenticeship theory. In a study of 245 first-year high school students who sequentially experienced three lesson formats (instructor voice, single-speaker TTS, dialogue TTS; content differed across sessions, limiting format/content separation), we conducted within-subject (Friedman test, N<=183) and repeated cross-sectional (Mann-Whitney U, N=229/206) analyses. TTS audio did not substantially degrade the learning exp
arXiv:2607.12149v1 Announce Type: new Abstract: Content moderation practices and governance paradigms are changing rapidly, as fewer human moderators are deployed as `experts' by social media companies in a centralized manner. Instead, the companies are focusing more on community approaches, relying on volunteers to provide accurate information and make correct decisions. In decentralized moderation, communities have always relied on volunteers, updated community guidelines, and internal discussions thereof. For both content moderation paradigms, Artificial Intelligence (AI) seems like it could help ease moderation burdens of time, mental health, and accuracy. One possible way to operationalize AI in content moderation is a `policy-as-prompt'' approach, where the policy is formulated as a natural-language prompt and then passed to a large language model (LLM). This model then aids in moderation tasks. In this paper, we briefly lay out the technical and governance properties of this app
arXiv:2607.11999v1 Announce Type: new Abstract: From maritime trade to commercial nuclear power, insurance has been the enabler of major economic and technological developments by pricing risk, limiting downside, and spreading best practices. The emerging AI agent economy, projected to handle trillions of dollars in transactions by 2030, looks to be the next such development. Yet insurers' exposure to AI agent risk currently sits largely unpriced across existing insurance lines; between this silent coverage and growing exclusions, coverage is not fit for purpose. Furthermore, insurability is trending the wrong way: AI agent capabilities appear to be outpacing reliability, leading to rising incident severity; concentration among a few foundation model providers threatens correlated losses; and traditional actuarial modeling will struggle to keep pace with a technology evolving as rapidly as frontier AI. This report argues that affirmative AI coverage with limits in the billions is achie
arXiv:2607.11895v1 Announce Type: new Abstract: AI scientist systems are beginning to automate parts of scientific research, but social science poses a distinct challenge: its objects of inquiry are not merely datasets or laboratory protocols, but integrated social processes involving situated participants, interaction contexts, interventions, and outcomes. Yet a critical link is missing: existing systems either assist isolated research tasks or simulate agents as experimental subjects, leaving the research workflow and simulated society decoupled. Here we introduce AgentSociety 2, an Integrated Research Environment for executable social science. It couples two roles of LLM agents in the same runtime: AI social scientists that coordinate literature grounding, hypothesis generation, experiment design, simulation execution, result interpretation, and manuscript drafting; and silicon participants that generate behavioral responses within configurable social environments. This dual-role de
arXiv:2607.11890v1 Announce Type: new Abstract: Open-ended surveys offer valuable insights, but they are notoriously difficult to analyze at scale. Building on previous work that employed traditional machine learning to classify text ("So Many Responses, So Little Time: A Machine-Learning Approach to Analyzing Open-Ended Survey Data") [1], this study investigates how different large language models (LLMs) understand and analyze NSSE open-ended survey responses. We focus on several cutting-edge LLMSs-OpenAI's GPT series, Twitter-roBERTa-base model, and Meta's LLaMA-and compare their performance to the previous machine learning models in tasks like sentiment analysis and thematic classification. Our research analysis assesses model agreement, classification accuracy, and interpretability of reasoning. The findings reveal that current LLMs routinely beat classic machine learning models in classification accuracy, particularly in understanding complex mood and theme patterns in student rep
The 1842 Fund launched a new startup called Suvi Health, which is developing an ambient AI platform that records hospital bedside conversations and turns them into a shared task list for patients and families. The company will pilot the tool with Mayo Clinic starting next month. The post New Startup Launches to Close the Bedside Communication Gap in Hospitals appeared first on MedCity News .
A new survey finds that employers are moving away from the Big Three PBMs while prioritizing healthcare affordability and transparency. The post Report: Employers’ Use of Big Three PBMs Falls in 2026 appeared first on MedCity News .
Deerfield Management formed Boulevard Bio with scientific co-founders Georg Schett and Frank Nestle. The startup’s lead program is in early clinical development for an autoimmune kidney disorder, and could offer dosing advantages over a recently approved Vera Therapeutics medicine and a Vertex Pharmaceuticals drug currently under FDA review. The post Startup Boulevard Bio Launches With Immune Reset Pioneer Schett as Co-Founder appeared first on MedCity News .
The percentage of physical robots in industrial settings that are designed to work alongside humans has more than quadrupled since 2017, according to data from the nonprofit International Federation of Robotics. To investigate how collaborative robots can effectively communicate with people in manufacturing, medical and other settings, a number of colleges and universities have launched research efforts that involve virtual reality (VR), which may also help provide students with valuable workplace skills and experience. Vision-based training can be a powerful way to tap into the speed,…
The EdTech Quality Collaborative, a group of six education-focused organizations, recently released a new tool for assessing and purchasing K–12 technology. The guide grew from an issue that touches every district in the nation: tech sprawl. The average district accesses nearly 3,000 distinct ed tech products in a year, but until now there was no K–12-specific framework for evaluating these tools. That lack of guidance — and the disconnect between procurement and actual classroom use — often leads to low tech adoption and wasted funds. The collaborative, made up of 1EdTech, CAST, CoSN,…
While organizations have invested heavily in expanding ambulatory networks, many are discovering that operational workflows have not evolved at the same pace. The result is a growing gap between demand and the systems designed to manage it. The post Ambulatory Growth Demands a New Operating Model appeared first on MedCity News .
As the largest single-district community college system in the nation, Maricopa Community Colleges educates more Arizona residents than its three large public universities combined. The system operates 10 main campuses across the Phoenix metropolitan area and more than 30 satellite sites, serving nearly 100,000 students and employing 14,000 faculty members and staff in the 2026 spring semester. Those superlatives and numbers are all familiar to Jamie Spradlin, MCC’s CISO. When he took his position in June 2024, his first order of business was to establish a cybersecurity program…
Article URL: https://adamexpostand.substack.com/p/simo-educacion-2026-a-practical-exhibitor Comments URL: https://news.ycombinator.com/item?id=49271263 Points: 2 # Comments: 0
Artificial intelligence is becoming increasingly embedded across higher education. As AI policies continue to evolve, faculty are gaining greater clarity and opportunity to responsibly use new tools, including AI-powered assistants, content generation tools, chatbots, and a growing number of AI-enabled applications. The post Higher education needs better AI experiences, not more AI tools appeared first on eCampus News .
Veterans and active-duty military members together make up about 5% of college students in the U.S. Educator and retired Major General Matt Smith shares tips for connecting with these students.
Artificial intelligence is becoming increasingly embedded across higher education. As AI policies continue to evolve, faculty are gaining greater clarity and opportunity to responsibly use new tools, including AI-powered assistants, content generation tools, chatbots, and a growing number of AI-enabled applications. The post Higher education needs better AI experiences, not more AI tools appeared first on eCampus News .
Revenue cycle vendor Unlimited Technology Systems disclosed a ransomware attack affecting 3.8 million patients — the second-largest healthcare data breach reported to HHS this year. The post Health IT Vendor’s Data Breach Exposes Nearly 4M Patient Records appeared first on MedCity News .
Article URL: https://www.newyorker.com/magazine/2026/08/17/hannah-arendt-life-of-the-mind-thomas-meyer-book-review-an-admirable-woman-arthur-cohen Comments URL: https://news.ycombinator.com/item?id=49267500 Points: 3 # Comments: 1
arXiv:2608.00144v2 Announce Type: replace-cross Abstract: Membership inference (MIA) on language models is usually summarised by aggregate ROC-AUC, but such evaluations are confounded: model-free blind baselines can separate members from non-members using surface text alone. Building on probabilistic discoverable extraction, we study black-box training-data leakage using N samples from p_theta(. | x), placing mean overlap, extreme-value overlap, and self-concentration on a common functional-estimation footing. On WikiMIA, a blind bag-of-words classifier reaches AUC 0.97 (TPR 0.90 at 5% FPR) while sampling adds nothing. On an IID Pile split (MIMIR), neither self-concentration nor gold-continuation recovery significantly exceeds a blind baseline in aggregate. Aggregate metrics hide the real harm: sampling verbatim-extracts training data for a tail of documents no blind attack can reach. On Pythia-6.9B, 16.6% of 500 Pile documents bearing a real identifier (83 documents; 21.3% of those be
arXiv:2607.19364v2 Announce Type: replace-cross Abstract: Activation steering adds a residual-stream direction at inference time, providing lightweight behavioral control without fine-tuning. Sparse autoencoders (SAEs) can make such interventions auditable by decomposing dense activations into an approximately monosemantic feature basis. We introduce SAE-StatSteer, a transparent, optimization-free pipeline. It first filters features through six reliability conditions, then ranks the survivors by an unweighted Borda consensus over three statistics, an $F$-test, KSG mutual information, and Cohen's $d$, and finally combines the selected SAE decoder rows using Cohen's-$d$ weights. We evaluate three Gemma-family models across four behavioral domains against seven dense or SAE-based baselines. Our quality-conditioned protocol requires attribute movement while preserving relevance, richness, and coherence. Raw success systematically overstates usable control because strong shifts often degrad
arXiv:2606.07943v2 Announce Type: replace-cross Abstract: Agent skills extend general-purpose agents, but their open format enables skill poisoning: a tampered skill can make an agent run an attacker's command while completing the user's legitimate task. Invocation alone is insufficient; the attack-specific action must complete while that task still passes its verifier. We therefore define Attack Success Rate (ASR) to require a postcondition-validated sandbox action and a passing task verifier in the same trial. Skill files expose a reliability-visibility trade-off between a preloaded but conspicuous YAML frontmatter block and a longer body, where arbitrary placement may be skipped or locally incongruent. We introduce Poise, a position-aware attack that uses context-aware generation to place exactly one benign-looking, command-bearing instruction at a structurally feasible body position. On the eligible Skill-Inject pool with codex+gpt-5.2, Poise achieves 89.3\% ASR, 28.0 points above
arXiv:2606.00671v2 Announce Type: replace-cross Abstract: We present AXIOM, a trust-first neuro-symbolic architecture for natural-language mathematical reasoning. Its language model is strictly a canonicalizer: it rewrites informal problem text into a narrow schema consumed by a deterministic Computer-Algebra-System (CAS) pipeline, which derives and verifies the answer or abstains as a first-class output. Routing follows a 1:1:1 alignment between problem-shape regex, schema-specific prompt, and closed-form CAS handler, with 4,783 such routes shipped, 71% of which answer without invoking the language model, and zero LOST_CORRECT regressions as a standing release gate. Because the answer is derived rather than generated, so is its explanation: every handler emits a step trace of the computation it performed, rendered as prose by a layer covering all 4,785 tasks that cannot narrate a step the handler did not take. We report two numbers and never fuse them. On the full 7-category MATH test
arXiv:2605.16411v3 Announce Type: replace-cross Abstract: Hallucination remains a fundamental challenge in vision-language models (VLMs), where autoregressive generation may produce linguistically plausible yet physically inconsistent or visually ungrounded responses due to likelihood maximization under joint probabilistic modeling. We propose a stage-wise preference optimization framework for hallucination reduction through targeted multimodal data construction. Rather than directly optimizing on generic instruction-following data, our approach progressively constructs hallucination-focused preference pairs near known failure boundaries. The framework emphasizes ambiguous spatial orientation, object relationships, OCR uncertainty, and adversarial false-premise training. Hallucinated negatives are generated through minimally perturbed yet visually inconsistent alternatives, enabling Direct Preference Optimization (DPO) to better separate grounded reasoning from plausible hallucination.
arXiv:2605.05629v4 Announce Type: replace-cross Abstract: We study the problem of learning generative models for discrete sequences in a continuous embedding space. Whereas prior approaches typically operate in Euclidean space or on the probability simplex, we instead work on the sphere $\mathbb S^{d-1}$. There the von Mises-Fisher (vMF) distribution induces a natural noise process and admits a closed-form conditional score. The conditional velocity is in general intractable. Exploiting the radial symmetry of the vMF density we reduce the continuity equation on $\mathbb S^{d-1}$ to a scalar ODE in the cosine similarity, whose unique bounded solution determines the velocity. The marginal velocity and marginal score on $(\mathbb S^{d-1})^L$ both decompose into posterior-weighted tangent sums that differ only by per-token scalar weights. This gives access to both ODE and predictor-corrector (PC) sampling. The posterior is the only learned object, trained by a cross-entropy loss. Experimen
arXiv:2604.16706v2 Announce Type: replace-cross Abstract: Automated evaluation of tool-using large language model (LLM) agents is widely assumed to be reliable, yet this assumption is rarely validated against human annotation. We present AgentProp-Bench, a diagnostic benchmark of 14,750 execution traces from thirteen LLM agents (nine proprietary, four open-weight) across four domains, and use it to audit three questions. First, substring-heuristic judging of agent outputs agrees with human annotation only at chance level (Cohen's kappa = 0.049 against each of two annotators), while a three-LLM ensemble reaches moderate agreement (kappa = 0.432) and a single GPT-4o-mini judge is in fact the strongest (kappa = 0.567); dual-annotator agreement is almost perfect (kappa = 0.835). Second, under validated judging a parameter-level error propagates to a wrong final answer with human-calibrated probability approximately 0.62, replicated across proprietary and open-weight models, and a model's a
arXiv:2509.25143v2 Announce Type: replace-cross Abstract: Existing medical reasoning benchmarks for vision-language models primarily focus on analyzing a patient's condition based on an image from a single visit. However, this setting deviates significantly from real-world clinical practice, where doctors typically refer to a patient's historical conditions to provide a comprehensive assessment by tracking their changes over time. In this paper, we introduce TEMMED-BENCH, a multi-task benchmark designed for analyzing changes in patients' conditions between different clinical visits, which challenges large vision-language models (LVLMs) to reason over temporal medical images. TEMMED-BENCH consists of a test set comprising three tasks - visual question-answering (VQA), report generation, and image-pair selection - and a supplementary knowledge corpus of over 17,000 instances. With TEMMED-BENCH, we conduct an evaluation of twelve LVLMs, comprising six proprietary and six open-source model
arXiv:2509.23452v2 Announce Type: replace-cross Abstract: Current text-to-image generation models, even state-of-the-art models, exhibit a significant performance gap when spatial expressions are described from non-camera perspectives. To address this limitation, we propose Frame of Reference-guided Spatial Adjustment in LLM-based Diffusion Editing (FoR-SALE), an extension of the Self-correcting LLM-controlled Diffusion (SLD). FoR-SALE first evaluates the alignment between a given text and an initially generated image, and then refines the image based on the expressed FoR in the spatial description. It employs vision modules to extract the spatial configuration of the generated image and simultaneously maps the spatial expression to a corresponding camera perspective. This unified perspective enables direct evaluation of alignment between language and vision. When misalignment is detected, the required editing operations are generated and applied. FoR-SALE introduces novel latent-space
arXiv:2509.23102v4 Announce Type: replace-cross Abstract: Reinforcement learning from human feedback (RLHF) has emerged as the standard paradigm for aligning large language models with human preferences. However, reward-based methods grounded in the Bradley-Terry assumption struggle to capture the nontransitivity and heterogeneity of real-world preferences. To address this, recent studies have reframed alignment as a two-player Nash game, giving rise to Nash learning from human feedback (NLHF). While this perspective has inspired algorithms such as INPO, ONPO, and EGPO that offer strong theoretical and empirical guarantees, they remain fundamentally restricted to two-player interactions, introducing a single-opponent bias that fails to capture the full complexity of realistic preference structures. This work introduces Multiplayer Nash Preference Optimization (MNPO), a novel framework that generalizes NLHF to the multiplayer regime. It formulates alignment as an n-player game, where ea
arXiv:2502.02068v3 Announce Type: replace-cross Abstract: This paper introduces RoSeMary, the first-of-its-kind ML/Crypto codesign watermarking framework that regulates LLM-generated code to avoid intellectual property rights violations and inappropriate misuse in software development. High-quality watermarks adhering to the detectability-fidelity-robustness tri-objective are limited due to codes' low-entropy nature. Watermark verification, however, often needs to reveal the signature and requires re-encoding new ones for code reuse, which potentially compromising the system's usability. To overcome these challenges, RoSeMary obtains high-quality watermarks by training the watermark insertion and extraction modules end-to-end to ensure (i) unaltered watermarked code functionality and (ii) enhanced detectability and robustness leveraging pre-trained CodeT5 as the insertion backbone to enlarge the code syntactic and variable rename transformation search space. In the deployment, RoSeMary
arXiv:2607.13205v2 Announce Type: replace Abstract: Attention-based KV cache eviction (H2O and its descendants) compresses the memory-constrained state of a long-context model by ranking tokens on accumulated attention mass, treated here as signal energy, and keeping the heaviest. On schema-dense input streams such as nested JSON, this score acts as a non-stationary filter that disproportionately retains noise: a non-content sink role (delimiters or whitespace) carries an order of magnitude more energy than any content role, and structural KEY tokens are over-retained at roughly 1.8x the rate of the answer-carrying VALUE tokens, collapsing exact-match accuracy from 88% to 0% at a 5% budget as the signal-to-noise ratio of the retained state degrades. A counterfactual experiment establishes that suppressing KEY tokens is the best deployable filter. Our retraining-free, role-conditional allocation over SnapKV's windowed score, governed by a single tuned hyperparameter, closes 63-98% of th
arXiv:2606.15821v2 Announce Type: replace Abstract: Recent advances in large language models (LLMs) have produced many specialized multimodal LLMs (MLLMs) that share common foundational LLMs, forming distinct model lineages. It remains unclear whether a fundamental behavioral link exists between the foundational LLMs and downstream variants. We investigate this question by quantifying head-level context-truthfulness scores. Across diverse LLM and MLLM lineages, including Vicuna-, Qwen2.5-, LLaMA2-, and Mistral-based models, we find that Truth Scores are strongly preserved within model families, even after instruction tuning or multimodal adaptation. We further show that this inheritance is consistent with attention-head weight preservation, and that context-truthful heads attend to query-relevant evidence. Building on this finding, we propose TruthProbe, a soft-gating strategy that amplifies context-truthful heads while preserving other head contributions. TruthProbe improves contextua
arXiv:2606.11470v2 Announce Type: replace Abstract: Reasoning has become central to how Large Language Models (LLMs) are evaluated and interpreted, spanning Chain-of-Thought (CoT), mathematical problem-solving, multi-hop question answering, code generation, retrieval-augmented reasoning, tool use, and multimodal decision-making. In this survey, we introduce the Periodic Table of LLM Reasoning, a framework organizing 300+ recent papers by reasoning paradigm, methodological mechanism, evaluation setting, and failure mode. We classify LLM reasoning into nine paradigms: Chain-of-Thought, Multi-Hop, Mathematical, Commonsense, Visual and Temporal, Code and Algorithmic, Retrieval-Augmented, Tool-Augmented or Agentic, and Reinforcement Learning-based reasoning. For each, we review approaches, including prompting, architectural interventions, supervised fine-tuning, verifier-guided inference, reward modeling, retrieval, tool interfaces, agentic workflows, and benchmark design. We argue that LLM
arXiv:2606.08348v2 Announce Type: replace Abstract: LLM agents increasingly rely on prompts, tools, memory, SOPs, skills, and harness feedback, yet current self-evolution pipelines often update these assets through heuristic reflection or raw success counts. Such updates are brittle when trajectories are sparse, expensive, and context-dependent. We introduce Bayesian-Agent, a native and cross-harness framework that treats reusable agent skills as Bayesian evidence objects. Bayesian-Agent records verified trajectories, maintains posterior beliefs over skill reliability and failure modes, and turns those beliefs into auditable skill actions and model-facing guardrails. This posterior view provides a finite-sample alternative to raw empirical-rate skill updates and frames prompt, context, and harness engineering as inference over the external decision environment. On RealFin-Bench, Bayesian skill evolution matches or improves the raw empirical-rate control and yields large gains on native
arXiv:2606.06960v2 Announce Type: replace Abstract: Experience-based self-evolution enables language-model agents to improve their behavior by accumulating and updating experience at test time, yet existing evaluations often assume recurring task patterns and explicit success signals. We introduce \textsc{FinEvolveBench}, a benchmark for self-evolving agents on low-repetition tasks with implicit rewards. The benchmark reconstructs a daily financial information stream over 31 Chinese A-share industry indices and aligns 177,324 public news articles with market observations. Researchers can define prediction horizons over this stream; we evaluate predictive market-sentiment factors against delayed market-adjusted returns after 10, 20, and 40 trading days. Unlike static benchmarks that score each prediction independently, \textsc{FinEvolveBench} interleaves new decisions with delayed outcomes from earlier ones, testing whether agents can convert noisy real-world feedback into reusable expe
arXiv:2606.03793v2 Announce Type: replace Abstract: Multimodal Large Language Models integrate visual perception into language reasoning, introducing a continuous attack surface susceptible to adversarial attacks. Prior work on MLLM robustness has focused largely on English-centric tasks, leaving multilingual behaviour unexplored. We address this gap through a systematic study of adversarial robustness and multimodal safety across 12 diverse languages, evaluating open-source MLLMs that acquire multilingual capability through instruction tuning. Gradient-based attacks reveal a transferable multilingual vulnerability: adversarial images optimized in one language continue to induce failure in others, demonstrating strong cross-lingual transferability. Multilingual safety further varies with how effectively a model retrieves or interprets harmful instructions. When harmful intent is issued through text, languages with stronger linguistic grounding more often elicit misuse-enabling response
arXiv:2605.17173v2 Announce Type: replace Abstract: Large language models exhibit safety degradation in non-English languages. Standard evaluation relies on Jailbreak Success Rate (JSR), which confounds several safety-driving factors into one, obscuring the specific cause(s) of safety failure. We introduce a latent variable model, a Multi-Group Item Response Theory (IRT) framework, that decouples language-agnostic safety robustness ($\theta$), intrinsic prompt hardness ($\beta$), global language processing difficulty ($\gamma$), and a prompt-specific cross-lingual safety gap ($\tau$). Using the MultiJail dataset, we evaluate the safety robustness of 61 model configurations across 5 closed-model families and 10 languages of varying resource, aggregating a dataset of 1.9 million responses. Exploratory Factor Analysis shows safety is primarily unidimensional: models refuse different harm types mainly through a shared mechanism. Contrary to the expected trend that safety degrades largely i
arXiv:2605.11533v3 Announce Type: replace Abstract: Routine clinical check-up reports combine laboratory measurements, physiological assessments, imaging findings and visually structured information, but rarely tell patients what to do next. Translating them into follow-up actions requires models to connect evidence across pages, tables and modalities, identify clinically relevant issues and communicate next steps without unsupported diagnostic or treatment claims. Yet this report-to-action capability remains poorly benchmarked. We introduce \textbf{C2A}, a dataset and benchmark for generating structured \textit{Action Cards} from multimodal check-up reports, together with \textbf{Checkup2Action}, a constrained workflow for the task. C2A contains 2,000 de-identified real-world reports covering physical examinations, laboratory tests, cardiovascular assessments and imaging evidence. Each card specifies one issue, its priority, recommended department, follow-up window, patient-facing exp
arXiv:2605.02815v2 Announce Type: replace Abstract: Text-to-SQL over large analytical databases requires navigating complex schemas, resolving ambiguous queries, and grounding decisions in actual data. Most current systems follow a fixed pipeline where schema elements are retrieved once upfront and the database is only revisited for post-hoc repair, limiting recovery from early mistakes. We present FlexSQL, a text-to-SQL agent whose core design principle is flexible database interaction: the agent can explore schema structure, inspect data values, and run verification queries at any point during reasoning. FlexSQL generates diverse execution plans to cover multiple query interpretations, implements each plan in either SQL or Python depending on the task, and uses a two-tiered repair mechanism that can backtrack from code-level errors to plan-level revisions. On Spider2-Snow, using gpt-oss-120b, FlexSQL achieves a 65.4\% score, outperforming strong open-source baselines that use stronge
arXiv:2604.25374v2 Announce Type: replace Abstract: Background: Dutch medical corpora are scarce, limiting NLP development. Methods: We translated English datasets, identified medical text in generic corpora, and extracted open Dutch medical resources. Results: The resulting corpus comprises +- 36 billion tokens across the medical domain in about 105 million documents, freely available on Hugging Face. Conclusion: This work establishes the first large-scale Dutch medical language corpus for pre-training and downstream NLP tasks.
arXiv:2604.23733v2 Announce Type: replace Abstract: Discourse comprehension in complex documents often involves continuously posing and resolving Questions Under Discussion (QUDs). While QUD frameworks have so far focused on text, scientific literature is inherently multimodal: figures convey discourse goals distinct from their textual counterparts, thus invoking implicit questions that the surrounding text answers. In scientific discovery, knowing the right questions to ask is as important as knowing how to answer them, yet this capability remains largely absent in current models. In this work, we extend QUD to multimodal discourse in scientific literature, targeting questions evoked by figures that are (1) inquisitive, i.e., not resolved in the prior context; (2) salient, i.e., relevant to the paper's research claims and addressed later in the paper; (3) grounded in visual insights. To benchmark model capability to generate such questions, we introduce MQUD, a dataset of 1,250 figure