Academic Intelligence · Curated Daily

Explore the Frontier of Global Academia

AcademicHub aggregates real-time literature from top journals and preprint platforms. Build your personal research radar and let large language models compile cross-disciplinary analysis briefings automatically.

01.
arXiv (CS.CL) 2026-06-24

Removing Noise, not Finding Gold: Quality Filtering for Large-Scale Pretraining

Large-scale models are pretrained on massive web-crawled datasets containing documents of mixed quality, making data filtering essential. A popular method is Classifier-based Quality Filtering (CQF), which trains a binary classifier to distinguish between pretraining data and a small, high-quality set. It assigns each pretraining document a quality score defined as the classifier's score and retains only the top-scoring ones. We provide an in-depth analysis of CQF. We show that while CQF improves downstream task performance, it does not necessarily enhance language modeling on the high-quality dataset. We explain this paradox by the fact that CQF implicitly filters the high-quality dataset as well. We further compare the behavior of models trained with CQF to those trained on synthetic data of increasing quality, obtained via random token permutations, and find starkly different trends. Our results challenge the view that CQF captures a meaningful notion of data quality.

02.
arXiv (CS.LG) 2026-06-12

Central Limit Theorems for Stochastic Gradient Descent Quantile Estimators

arXiv:2503.02178v3 Announce Type: replace-cross Abstract: This paper develops asymptotic theory for quantile estimation via stochastic gradient descent (SGD) with a constant learning rate. The quantile loss function is neither smooth nor strongly convex. Beyond conventional perspectives and techniques, we view quantile SGD iteration as an irreducible, periodic, and positive recurrent Markov chain, which cyclically converges to its unique stationary distribution regardless of the arbitrarily fixed initialization. To derive the exact form of the stationary distribution, we analyze the structure of its characteristic function by exploiting the stationary equation. We also derive tight bounds for its moment generating function (MGF) and tail probabilities. Synthesizing the aforementioned approaches, we prove that the centered and standardized stationary distribution converges to a Gaussian distribution as the learning rate $\eta\rightarrow0$. This finding provides the first central limit theorem (CLT)-type theoretical guarantees for the quantile SGD estimator with constant learning rates. We further propose a recursive algorithm to construct confidence intervals of the estimators with statistical guarantees. Numerical studies demonstrate the effective finite-sample performance of the online estimator and inference procedure. The theoretical tools developed in this study are of independent interest for investigating general SGD algorithms formulated as Markov chains, particularly in non-strongly convex and non-smooth settings.

03.
arXiv (CS.CL) 2026-06-24

Knowledge-Graph Grounding Helps LLMs Only for Out-of-Training Knowledge: A Controlled Study on Clinical Question Answering

A recent Nature Medicine study reports that general-purpose frontier LLMs outperform specialized retrieval-augmented clinical tools on medical benchmarks, and that retrieval can hurt strong models. We ask the natural follow-up: does structured knowledge-graph (KG) grounding change this, and when does grounding help at all? We contribute two results. First, a reproduction: the study's headline HealthBench score (~88) is the Consensus variant, not full HealthBench, where frontier models and ideal completions both score ~46-47 under a physician-calibrated grader (agreement 82.5%); we reproduce GPT-5.2 Consensus =90.9 and flag a score-deflating grader bug. Second, a knowledge-boundary result. Using a graph+vector engine (samyama-graph) over the public biomedical KG PrimeKG, neither naive triple retrieval nor an agentic natural-language-to-Cypher loop (82% successful queries) improves MedQA across a weak-to-strong model ladder (all |Delta|

04.
arXiv (CS.LG) 2026-06-16

Not All Retrievals are Useful: Cross-Attention for Input-Aware RAG in Time Series Forecasting

arXiv:2603.14709v2 Announce Type: replace Abstract: Retrieval-augmented generation (RAG) enhances zero-shot time series (TS) forecasting by leveraging external knowledge bases, yet existing approaches overlook input-level relevance when fusing retrieved samples with the query. We argue that not all retrievals are equally useful, and irrelevant ones can degrade performance. To this end, we propose Cross-RAG, a zero-shot RAG-based forecasting framework that selectively attends to query-relevant retrieved samples via query–retrieval cross-attention. By modeling input-level relevance between the query and retrieved samples, Cross-RAG jointly incorporates three sources of information: 1) the query itself, 2) the retrieved samples, and 3) their relational interactions. In particular, this input-aware design enables Cross-RAG to remain stable as the number of retrieved samples $k$ grows, whereas prior methods without cross-attention require careful $k$ tuning to avoid degradation from irrelevant retrievals. Extensive experiments demonstrate that Cross-RAG consistently improves zero-shot forecasting performance across multiple TSFM backbones and various RAG methods, with additional analyses confirming its effectiveness across various retrieval scenarios. Code is available at https://github.com/seunghan96/cross-rag/.

05.
arXiv (CS.LG) 2026-06-25

Robust Linear Predictions: Analyses of Uniform Concentration, Fast Rates and Model Misspecification

arXiv:2201.01973v3 Announce Type: replace-cross Abstract: The problem of linear predictions has been extensively studied for the past century under pretty generalized frameworks. Recent advances in the robust statistics literature allow us to analyze robust versions of classical linear models through the prism of Median of Means (MoM). Combining these approaches in a piecemeal way might lead to ad-hoc procedures, and the restricted theoretical conclusions that underpin each individual contribution may no longer be valid. To meet these challenges coherently, in this study, we offer a unified robust framework that includes a broad variety of linear prediction problems on a Hilbert space, coupled with a generic class of loss functions. Notably, we do not require any assumptions on the distribution of the outlying data points ($\mathcal{O}$) nor the compactness of the support of the inlying ones ($\mathcal{I}$). Under mild conditions on the dual norm, we show that for misspecification level $\epsilon$, these estimators achieve an error rate of $O(\max\left\{|\mathcal{O}|^{1/2}n^{-1/2}, |\mathcal{I}|^{1/2}n^{-1} \right\}+\epsilon)$, matching the best-known rates in literature. This rate is slightly slower than the classical rates of $O(n^{-1/2})$, indicating that we need to pay a price in terms of error rates to obtain robust estimates. Additionally, we show that this rate can be improved to achieve so-called "fast rates" under additional assumptions.

06.
bioRxiv (Bioinfo) 2026-06-24

InVitroGap: an open-source tool for automated quantification of wound closure in the in vitro scratch assay

Abstract Background and Objective: Scratch assays are widely used to study wound closure in vitro, but quantitative image analysis remains constrained by manual variability, proprietary workflows, and tools requiring programming expertise. We developed InVitroGap, a Python-based application with a browser-accessible interface for automated quantification of scratch assay closure from sequential microscopy images. Methods: RCC-ER and Renca cells were seeded in 96-well ImageLock plates and scratched using a WoundMaker device for uniform linear wounds or a 200 uL pipette tip for crisscross wounds. Phase-contrast time-lapse images acquired at 0, 24, and 48 h with an IncuCyte SX5 system were independently analyzed using IncuCyte 2023A Rev2 and InVitroGap. The InVitroGap pipeline combines Gaussian smoothing, gradient-based texture mapping, adaptive percentile thresholding, and morphological post-processing to quantify wound confluence and relative wound density (RWD). Agreement was evaluated using paired comparisons, Pearson and Spearman correlations, Bland-Altman analysis, and mean absolute error (MAE). Results: InVitroGap measurements closely tracked IncuCyte outputs across both cell lines, with no significant between-method differences (p > 0.05), strong pooled correlations (R square = 0.964 for RWD; R square = 0.983 for wound confluence), and small mean biases (absolute bias [≤] 1.64%). The tool successfully processed crisscross wounds from brightfield image series, and a complete four-timepoint series was analyzed in approximately 10 seconds, with robust performance across distinct cell morphologies and wound geometries. Conclusions: InVitroGap provides a transparent, computationally efficient, and platform-independent alternative for scratch assay analysis, delivering performance comparable to commercial systems while remaining freely accessible at https://invitrogap.vercel.app/.

07.
arXiv (quant-ph) 2026-06-15

Multiple-time Quantum Imaginary Time Evolution

arXiv:2512.10875v2 Announce Type: replace Abstract: Quantum Imaginary-Time Evolution (QITE) is a powerful method for preparing ground states on quantum hardware. However, executing QITE has costly measurement budgets for general Hamiltonians. Both fidelity and computational cost are strongly dependent on the definition of suitable local domains and Hamiltonian partitions. In this work, we introduce the Multiple-Time QITE algorithm (MT-QITE). We show how using more than one imaginary time substantially improves the fidelity of the resulting ground state as well as the measurement overhead with respect to the previously published QITE algorithm, while preserving its deterministic character and its independence from ad hoc ansatze. Moreover, unlike QITE and other QITE-based algorithms, MT-QITE is parallelizable, and we show that even in Hamiltonians with non-local interactions, partitioning may entail a computational advantage.

08.
arXiv (CS.LG) 2026-06-17

Learn from Your Mistakes: Self-Correcting Masked Diffusion Models

arXiv:2602.11590v3 Announce Type: replace Abstract: Masked diffusion models (MDMs) have emerged as a promising alternative to autoregressive models, enabling parallel token generation while achieving competitive performance. Despite these advantages, MDMs face a fundamental limitation: once tokens are unmasked, they remain fixed, leading to error accumulation and ultimately degrading sample quality. We address this by proposing a framework that trains a model to perform both unmasking and correction. By reusing outputs from the MDM denoising network as inputs for corrector training, we train a model to recover from potential mistakes. During generation we apply additional corrective refinement steps between unmasking ones in order to change decoded tokens and improve outputs. We name our training and sampling method Progressive Self-Correction (ProSeCo) for its unique ability to iteratively refine an entire sequence, including already generated tokens. We conduct extensive experimental validation across multiple conditional and unconditional tasks, demonstrating that \method~yields better quality-efficiency trade-offs (up to ~4x faster sampling) and enables inference-time compute scaling to further increase sample quality beyond standard MDMs (up to ~1.2x improvement on benchmarks).

09.
arXiv (CS.LG) 2026-06-16

Time-Varying Audio Effect Modeling by End-to-End Adversarial Training

arXiv:2512.15313v2 Announce Type: replace-cross Abstract: Deep learning has become a standard approach for the modeling of audio effects, yet strictly black-box modeling remains problematic for time-varying systems. Unlike time-invariant effects, training models on devices with internal modulation typically requires the recording or extraction of control signals to ensure the time-alignment required by standard loss functions. This paper introduces a Generative Adversarial Network (GAN) framework to model such effects using only input-output audio recordings, without requiring a modulation signal extraction. We propose a convolutional-recurrent architecture trained via a two-stage strategy: an initial adversarial phase allows the model to learn the distribution of the modulation behavior without strict phase constraints, followed by a supervised fine-tuning phase where a State Prediction Network (SPN) estimates the initial internal states required to synchronize the model with the target. Additionally, a new metric based on chirp-train signals is developed to quantify modulation accuracy. Experiments modeling a vintage hardware phaser demonstrate the method's ability to capture time-varying dynamics in a fully black-box context.

10.
medRxiv (Medicine) 2026-06-15

Routine use of oral iron for people with heart failure and iron deficiency in primary care; retrospective cohort study

Aims: Iron deficiency is common among people with heart failure and associated with morbidity and mortality. While intravenous iron improves clinical outcomes, oral iron continues to be prescribed in routine practice despite limited evidence of benefit. Methods: We completed a retrospective primary care cohort study (2016 to 2021) to investigate the proportion of people with an incident diagnosis of heart failure who had iron deficiency identified (defined as ferritin

11.
arXiv (quant-ph) 2026-06-11

Fermions are fundamentally more nonlocal than Bosons

arXiv:2606.12363v1 Announce Type: new Abstract: Bell's theorem shows that entangled quantum particles can exhibit correlations that classical particles cannot reproduce without an additional nonlocal resource, such as communication. In this sense, quantum particles are fundamentally more nonlocal than classical ones, and entanglement becomes unavoidable in physics. Here we prove the analogous result within quantum theory itself: indistinguishable fermions transmitted through a quantum network can generate correlations that distinguishable particles or indistinguishable bosons cannot reproduce without additional communication. In the same sense, fermions are fundamentally more nonlocal than bosons or distinguishable particles, motivating fermionic anticommutation and indistinguishability as unavoidable operational resources. Our result further implies that fermions can strictly surpass all qubit-based protocols for certain distributed computing tasks, demonstrating that a complete understanding of information processing requires going beyond qubits to fermionic information carriers - febits.

12.
arXiv (CS.AI) 2026-06-15

History of the Muddy Children Puzzle

arXiv:2606.13703v1 Announce Type: new Abstract: The Muddy Children Puzzle is a puzzle about knowledge and ignorance that has been inspiring for the development of epistemic logic. Who came up with it first? This is unclear. We trace the origin of the Muddy Children Puzzle through logical and literary publications over the past two centuries. The puzzle inspired a numerous variations such as involving numbers or coloured hats. We also present a novel hats puzzle involving self-reference.

13.
arXiv (CS.AI) 2026-06-16

Fine-Tuning a 7B Advisor on Free-Tier GPUs: An Adapter-Handoff Recipe and a Synthetic-Data Reliability Caution

arXiv:2504.15610v4 Announce Type: replace Abstract: Fine-tuning a 7B language model for specialized advising is attractive in resource-constrained settings, but multi-epoch runs routinely exceed the wall-clock limits of the free-tier GPUs (Kaggle, Colab) such users rely on. We report two things. First, a practical recipe: a three-epoch QLoRA fine-tune of Mistral-7B-Instruct-v0.3 (4-bit NF4, LoRA rank 16, via Unsloth) completed across two free-tier 16 GB GPUs (Tesla P100 then T4) by checkpointing only the small LoRA adapter (41.9M parameters) and resuming on the second machine. Adapter-only handoff is sufficient – optimizer and scheduler state need not be transferred – so the binding constraint is per-step VRAM and per-session wall-clock, not aggregate compute. Second, and more importantly, an honest evaluation that returns a cautionary result. On a blind held-out comparison against the un-fine-tuned base model, the fine-tuned model scored higher on similarity to the synthetic training distribution (BERTScore F1 +0.063, a fidelity not quality signal) but lower on advising quality: a blind LLM-as-judge preferred the base model on 46% of prompts versus 18%, and a source-verified factuality audit found four confident errors from the fine-tuned model on policy-sensitive topics against zero for the base. Auditing the training data with the same method, we find this is not a fine-tuning artifact: each audited error is already present in the Gemini-generated training answers, and a random-sample audit finds verifiable errors in a sizable fraction of responses (28-40%; single-judge, n=40). The data is therefore sufficient to account for the errors, which we attribute to the synthetic-data pipeline rather than the adapter-handoff method. We release the dataset, adapter, cross-GPU notebooks, and full evaluation harness so every result reproduces on a single 16 GB GPU.

14.
arXiv (CS.CL) 2026-06-25

Error-Aware TF-IDF Retrieval-Augmented Generation for ASR Error Correction

End-to-end automatic speech recognition systems frequently hallucinate rare entities and domain-specific terms, especially in low-resource languages. While retrieval-augmented generation frameworks can mitigate these errors using large language models, current architectures face significant challenges. They either rely on standard sparse retrieval that ignores phonetic misrecognitions or utilize heavyweight cross-modal embeddings that introduce high latency. This letter proposes a highly efficient, purely lexical error-aware framework designed to explicitly resolve phonetic and loop hallucinations. Our approach integrates a symmetric text normalization module with a novel error-aware term frequency-inverse document frequency algorithm. By constructing a sparse diagonal penalty matrix based on historical errors, the retriever mathematically prioritizes corrective documents containing specific high-risk misrecognitions. Evaluated on the Persian subset of the FLEURS dataset, our method increased the error-aware hit rate from 53.7% to 90.9%. In end-to-end evaluations, the integrated framework reduced the final word error rate from 23.06% to 18.83%, achieving significant accuracy gains with near-zero inference latency.

15.
arXiv (CS.CL) 2026-06-19

The Almost Intelligent Revolution: Options for Scaling Up Deliberation and Empowering People with AI

Authors:

The increasing prominence of Large Language Models (LLMs) in public discourse presents both opportunities and challenges for democratic deliberation. While red teaming strategies help mitigate specific risks, broader concerns persist regarding linguistic constraints, biases, and the sycophantic tendencies of LLMs. This chapter explores how LLMs can be used to significantly scale up and democratise deliberation, particularly in fostering inclusivity and empowering traditionally marginalised groups. Drawing on concepts from Systemic-Functional Linguistics, the chapter examines how variations across language users (for example, with respect to socio-demographic groups) and across language use (for example, with respect to communicative functions) shape participation in AI-supported deliberation. The chapter presents AI-driven deliberation studies and assesses their potential to scaffold argumentation, enhance access, and reduce the influence of exclusionary linguistic norms and biases which are embedded in prestigious registers. At the same time, the chapter cautions against both overclaiming, which leads to unrealistic expectations, and underclaiming, which risks missed opportunities for AI-assisted engagement. The chapter concludes by identifying future research directions to maximise the democratic potential of AI-assisted participation while embedding ethical safeguards to counteract the reproduction of linguistic inequalities.

16.
arXiv (CS.LG) 2026-06-17

When the Next Step Is Not One Step: Distribution-Aware Execution Modeling for Concurrent Go Programs

arXiv:2606.17508v1 Announce Type: new Abstract: Training a model to predict the next step in a concurrent program is harder than it looks: two runs of the same program from the same trace prefix can produce different next events, both valid, because the scheduler is nondeterministic. A model trained against a single label is learning to guess one outcome of a random process. We turn this around and use the nondeterminism as a training signal. We run each program many times, aggregate the observed next events into an empirical distribution, and fine-tune a 7B model to match that distribution with a KL objective. On 798 held-out predictions drawn from real production Go bugs (CockroachDB, Kubernetes, gRPC, etcd), fine-tuning on fewer than a thousand traces reaches 36.2% accuracy, ahead of Gemini 3.5 Flash used zero-shot (34.8%) and the same model without fine-tuning (28.6%). Distribution training matches cross-entropy on accuracy (35.8% vs. 36.2%) while reducing Expected Calibration Error from 0.205 to 0.169. We also derive a formal goroutine-leak signature for a class of select-blocked goroutines where P(GoUnblock)=0 holds by scheduler semantics, not by learning. We release the dataset, trained adapters, and all tooling.

17.
medRxiv (Medicine) 2026-06-22

MinderCare: protocol for a mixed-methods evaluation of a digitally enabled dementia care service.

Introduction and aims Dementia is a growing public health challenge affecting millions of people worldwide. It is a progressive condition that increases the risk of infections, falls, hospital admissions, dependence in activities of daily living, safety issues such as wandering, care home transfers, and death. New ways of supporting people living with dementia (PLWD) at home are urgently needed. We describe the MinderCare study which evaluates a digitally enabled care model that integrates low-burden sensor-based remote monitoring within a nurse-led clinical service. Methods and analysis In this mixed-methods study, we will recruit 100 people with confirmed or suspected dementia living at home and deploy the Minder remote monitoring system for at least 12 months. A detailed characterisation of the cohort will be obtained, including cognition, frailty, participant and carer wellbeing, functioning, and quality of life. The feasibility, acceptability, sustainability, and resource requirements of the service will also be assessed. Low-cost sensors provide information about behaviour, environment and physiology from the home. Machine-learning algorithms have been used to develop digital biomarkers of infection, sleep, night-time behaviours, daily activities and routines, and the effects of clinical events and treatment. These will be assessed through clinical reports of sensor-derived data that include anomaly alerts provided to the clinical teams. Algorithms will be assessed for their clinical utility and acceptability. The comparative-effectiveness component will be designed as a target trial emulation using linked electronic health-record data to construct a time-indexed external usual-care control cohort. The primary comparative outcome will be Days Alive and Out of Hospital (DAOH) over 12 months from the activation-index date, with healthcare utilisation, costs, institutionalisation and mortality assessed as secondary outcomes. DAOH and estimated MinderCare effects will also be examined across prespecified strata of baseline inpatient utilisation. Ethics and dissemination Ethical approval has been granted by the North East Newcastle and North Tyneside 2 Research Ethics Committee, and the study has received confirmation of capacity and capability by the Imperial College Healthcare NHS Trust. Study findings will be disseminated to patients, health and social care professionals, and policymakers through peer-reviewed publications and conference presentations. Study registration number: ISRCTN14997677 and NIHR portfolio CPMSID 63023.

18.
arXiv (CS.CV) 2026-06-25

H-Adapter: Pose-Robust Hairstyle Transfer via Attention-Derived, Source-Aligned Hair Masks

Hairstyle transfer has practical applications such as virtual try-on, yet remains challenging when the source and reference exhibit large head-pose discrepancies. We propose H-Adapter, which improves pose robustness by training with a region-specific loss that disentangles hair and non-hair objectives and thereby induces spatially disentangled cross-attention, from which a source-aligned hair edit mask is derived to guide diffusion-based inpainting. Experiments on pose-agnostic and pose-different subsets demonstrate strong quantitative results, including the best FID, $\mathrm{FID}_{\mathrm{CLIP}}$, and CLIP-I under pose differences, while maintaining competitive non-hair preservation and improving qualitative fidelity to fine-grained reference hairstyle details. Beyond source-conditioned transfer, H-Adapter supports practical extensions including text-to-image generation, auxiliary prompt-based hair color control, and compatibility with an identity-preserving IP-Adapter variant. We also introduce a VLM-as-a-judge protocol and observe consistent gains in hairstyle faithfulness, non-hair preservation, and artifact quality.

19.
arXiv (math.PR) 2026-06-17

Moment generating function of the tacnode process

Authors:

arXiv:2606.17771v1 Announce Type: cross Abstract: The tacnode process is a universal determinantal point process arising in non-intersecting particle systems and random tiling models. In this paper, we study the generating function for the counting functions of the tacnode process on a union of $m$ intervals, $m\in\mathbb{N}^{+}$. Our first result provides an integral representation for the $m$-point generating function in terms of the Hamiltonian governing a system of $8m+4$ coupled differential equations. Combined with several differential identities for this Hamiltonian, the representation yields the large gap asymptotics, up to and including the constant term. As further applications, we obtain asymptotic formulae for the expectations, variances, and covariances of the counting functions, and establish a central limit theorem for their joint fluctuations. These results extend the previously known $1$-point theory for the tacnode process to the multi-interval setting with multiple discontinuities.

20.
arXiv (CS.AI) 2026-06-17

OmniRetarget: Interaction-Preserving Data Generation for Humanoid Whole-Body Loco-Manipulation and Scene Interaction

arXiv:2509.26633v3 Announce Type: replace-cross Abstract: A dominant paradigm for teaching humanoid robots complex skills is to retarget human motions as kinematic references to train reinforcement learning (RL) policies. However, existing retargeting pipelines often struggle with the significant embodiment gap between humans and robots, producing physically implausible artifacts like foot-skating and penetration. More importantly, common retargeting methods neglect the rich human-object and human-environment interactions essential for expressive locomotion and loco-manipulation. To address this, we introduce OmniRetarget, an interaction-preserving data generation engine based on an interaction mesh that explicitly models and preserves the crucial spatial and contact relationships between an agent, the terrain, and manipulated objects. By minimizing the Laplacian deformation between the human and robot meshes while enforcing kinematic constraints, OmniRetarget generates kinematically feasible trajectories. Moreover, preserving task-relevant interactions enables efficient data augmentation, from a single demonstration to different robot embodiments, terrains, and object configurations. We comprehensively evaluate OmniRetarget by retargeting motions from OMOMO, LAFAN1, and our in-house MoCap datasets, generating over 8-hour trajectories that achieve better kinematic constraint satisfaction and contact preservation than widely used baselines. Such high-quality data enables proprioceptive RL policies to successfully execute long-horizon (up to 30 seconds) parkour and loco-manipulation skills on a Unitree G1 humanoid, trained with only 5 reward terms and simple domain randomization shared by all tasks, without any learning curriculum.

21.
arXiv (CS.AI) 2026-06-16

DeepRoot: A KG-Coordinated Multi-Agent System for Therapeutic Reasoning over Historical Medical Texts

arXiv:2606.15931v1 Announce Type: cross Abstract: Historical medical archives and traditional medicines hold immense potential for drug discovery and remain a primary source for current drug development. However, pre-ontological prose and idiosyncratic taxonomies prevent the standardization and medical modernization of the data for use in current biomedical pipelines. Furthermore, no existing LLM agent system, whether tool-calling, retrieval-augmented, or agentic deep-research, can convert such text into verifiable drug-discovery leads at scale. We close this gap with DeepRoot, a multi-agent LLM system that jointly builds and utilizes a verified knowledge graph, showing that grounding and reasoning – often conflated – are separable axes the system can compose for therapeutic reasoning. Applied to the Shen Nong Ben Cao Jing, DeepRoot recovers $10$ of $21$ held-out compound-disease treatment pairs at R@$20$ ($47.6\%$ vs $4.8\%$ for a raw corpus LLM and $\sim\!2.4\%$ random) and dominates an LLM-as-judge audit for reasoning quality over baseline LLMs and LLMs with direct tool-call access to the same APIs DeepRoot itself queries. Tool-using LLMs hallucinate evidence on $87\%$ of claims, versus 7-10% for DeepRoot. Graph-only inference hallucinates $0\%$ but ranks lowest on reasoning coherence; DeepRoot KG+LLM is the only condition to win on both axes, pointing toward a route for systematic mining and repurposing of historical medical knowledge.

22.
arXiv (CS.AI) 2026-06-17

Breaking the Code: Security Assessment of AI Code Agents Through Systematic Jailbreaking Attacks

arXiv:2510.01359v2 Announce Type: replace-cross Abstract: Code-capable large language model (LLM) agents are embedded in software engineering workflows where they can read, write, and execute code, raising "jailbreak" stakes beyond text-only settings. Prior evaluations emphasize refusal or harmful-text detection, leaving open whether agents compile and run malicious programs. We present JAWS-Bench (Jailbreaks Across WorkSpaces), a benchmark spanning three escalating workspace regimes mirroring attacker capability: empty (JAWS-0), single-file (JAWS-1), and multi-file (JAWS-M). We pair this with a hierarchical, executable-aware Judge Framework that tests (i) compliance, (ii) attack success, (iii) syntactic correctness, and (iv) runtime executability, to measure deployable harm. Across seven LLM backends from five families, prompt-only attacks in JAWS-0 achieve 61% compliance; 58% are harmful, 52% parse, and 27% run end-to-end. In JAWS-1, compliance reaches ~100% for stronger models with a mean ASR (Attack Success Rate) ~71%; JAWS-M raises mean ASR to ~75%, with 32% runnable attack code. Wrapping an LLM in an agent increases ASR by 1.6$\times$, by overturning initial refusals during planning and tool use. Similar trends hold for OpenHands, SWE-Agent, and OpenAI Codex, suggesting our JAWS-Bench is agent-agnostic. Category analyses identify which attack classes are most vulnerable and deployable, motivating execution-aware defenses and refusal-preserving agent designs.

23.
arXiv (CS.AI) 2026-06-24

HOLMES: Evaluating Higher-Order Logical Reasoning in LLMs

arXiv:2606.23238v2 Announce Type: replace Abstract: Logical reasoning is essential for reliable AI, yet existing benchmarks are largely first-order-logic-centric, focusing on object-level deduction over fixed predicates. This misses many realistic scenarios where models must reason over rules, predicates, functions, constraints, and decision procedures themselves. We introduce HOLMES (Higher-Order Logic Meets real-world Explainable Symbolic reasoning), the first real-world benchmark for higher-order symbolic reasoning in LLMs, containing 1379 instances. Built on higher-order logic, HOLMES pairs natural-language problems with HOL formalizations, ground-truth answers, verifiable reasoning traces, and fine-grained controllable reasoning factors across law and finance. Experiments show that current LLMs still struggle on HOLMES, with an average accuracy of only 50.64% and the best model reaching 59.54%. Our analyses further reveal that high final-answer accuracy can mask shortcut reasoning in conflict-resolution settings, while performance drops sharply under scope-conditioned and compositional reasoning. These findings identify higher-order symbolic reasoning as a key bottleneck for building reliable and verifiable LLMs. The project code and dataset are publicly available at https://github.com/wuyucheng2002/HOLMES.

24.
arXiv (CS.CV) 2026-06-17

MLLMs Get It Right, Then Get It Wrong: Tracing and Correcting Late-Layer Textual Bias

When vision contradicts text, multimodal large language models (MLLMs) consistently favor text, even when images provide clear evidence otherwise. This bias poses risks for applications requiring visual grounding, yet its cause remains unclear. In this paper, we uncover a surprising finding: models often get it right initially, forming correct vision-based predictions in their intermediate layers, before changing their minds and favoring text in the final output. We call this "late-layer textual override". The visual information is encoded, it simply does not survive to the output. More intriguingly, we find that how predictions change reveals whether they're correct: 85% of failures shift toward text, while 89% of successes shift toward vision. This directional signature enables a simple but powerful intervention: when we detect a confident visual prediction being suppressed, we restore it. We propose CALRD (Conflict-Aware Layer Reference Decoding), a training-free method that recovers overridden predictions at inference time. Experiments across five MLLMs of varying architectures demonstrate up to 9.4% absolute improvements on conflict benchmarks while largely preserving standard performance, without training or external knowledge. It recovers what the model already knew but failed to preserve.

25.
bioRxiv (Bioinfo) 2026-06-23

CellOS: Learning a World Model of Cellular State through Joint Embedding Prediction

Foundation models learned from single-cell transcriptomes are central to the prospect of AI virtual cell that can represent, query and predict cellular state. However, most current single-cell foundation models learn from a single view of gene expression and are optimized primarily through reconstruction or next-token prediction. As a result, they capture expression abundance but can-not explicitly reconcile complementary views of cellular state. Here we present CellOS, a multi-view foundation model that learns cellular representations from paired expression and perception views. CellOS integrates complementary views through a scalable three-stage training strategy that combines causal cell-sentence language modelling, function-preserving dense-to-mixture-of-experts expansion and latent-space alignment via an LLM-JEPA objective. Using this framework, we trained a 12-billion-parameter model on 390.5 million single-cell transcriptomes. Across diverse benchmarks spanning cell-state annotation, batch integration and perturbation-response prediction, CellOS consistently outperformed state-of-the-art single-cell foundation models in cell-state annotation and perturbation-response prediction while preserving robust batch integration. Together, these results suggest that predictive alignment between complementary cellular views provides a scalable path toward representation-centric cellular world models and transferable AI virtual cells.