Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
Nature (Science) 2026-06-10

Measurement of reactor neutrino oscillation with the first JUNO data

Neutrino oscillations (see refs. 1,2 and references therein), a quantum effect manifesting at macroscopic scales, are governed by lepton flavour mixing angles and neutrino mass-squared differences3 that are fundamental parameters of particle physics, representing phenomena beyond the Standard Model. Precision measurements of these parameters are essential for testing the completeness of the three-flavour framework, determining the mass ordering of neutrinos and probing possible new physics. The Jiangmen Underground Neutrino Observatory (JUNO)4 is a 20-ktonne liquid-scintillator detector located 52.5 km from multiple reactor cores, designed to resolve the interference pattern of reactor neutrinos with sub-percent precision5,6. Here we report, using the first 59.1 days of data collected since detector completion in August 2025, the first simultaneous high-precision determination of two neutrino oscillation parameters, $${\sin }^{2}{\theta }_{12}=0.3092\,\pm \,0.0087$$ and $$\Delta {m}_{21}^{2}=(7.50\,\pm \,0.12)\times 1{0}^{-5}\,{\mathrm{eV}}^{2}$$ for the normal mass ordering scenario, improving the precision by a factor of 1.6 relative to the combination of all previous measurements. These results advance the basic understanding of neutrinos, validate the design of the detector and indicate the readiness of JUNO for resolving the neutrino mass ordering with a larger dataset. The rapid achievement with a short exposure highlights the potential of JUNO to push the frontiers of precision neutrino physics and paves the way for its broad scientific programme. The first data of the Jiangmen Underground Neutrino Observatory deliver high-precision neutrino oscillation parameters, improving measurements and demonstrating readiness to determine neutrino mass ordering.

02.
arXiv (CS.CV) 2026-06-11

Cross-Modal Benchmarking for Robotic Perception in Natural Environments

Natural environments present a complex challenge to robotics perception systems. Current models, particularly vision foundation models, are largely trained on structured, urban environments leading to weaknesses in their perception for field robotics tasks. We showcase the limitations of current models using our recently released WildCross benchmark, a new cross-modal benchmark for place recognition and metric depth estimation in large-scale natural environments. WildCross comprises over 476K sequential RGB frames with semi-dense depth and surface normal annotations, each aligned with accurate 6DoF pose and synchronized dense lidar submaps. In this work, we provide an expanded analysis of the benchmark results from the recent WildCross benchmark, with particular emphasis on expanded metric depth estimation experiments. Access to the code repository and dataset for this work can be found at https://csiro-robotics.github.io/WildCross.

03.
arXiv (CS.CV) 2026-06-16

Selective Synergistic Learning for Video Object-Centric Learning

Typical video object-centric learning (VOCL) approaches employ slot-based frameworks that rely on reconstruction-driven encoder-decoder architectures, where learning is mediated by two spatial maps: attention maps from the encoder and object maps from the decoder. As these two distinct maps exhibit different properties, a recent dense alignment strategy attempted to reconcile this discrepancy by enforcing agreement across all spatio-temporal patches via contrastive learning. However, this indiscriminate alignment inadvertently propagates the inherent weaknesses of each module, such as noisy encoder predictions and blurred decoder boundaries. Moreover, computing dense similarities across all pairs incurs a computational cost quadratic in the total number of spatio-temporal patches, severely limiting scalability. Motivated by this, we propose Selective Synergistic Learning (SSync). Instead of exhaustive patch-to-patch alignment, SSync prevents error propagation by selectively distilling only the most reliable cues: leveraging the encoder strictly for boundary refinement and the decoder for interior denoising. This is realized via a pseudo-labeling with linear complexity, eliminating the need for quadratic spatial comparisons. Also, to prevent the reinforcement of architectural biases like slot redundancy, we introduce a transitive pseudo-label merging that consolidates overlapping slots based on spatio-temporal activation consistency. Extensive studies demonstrate that SSync improves decomposition quality and serves as a versatile, plug-and-play module while also exhibiting exceptional robustness to slot configurations. Code is available at github.com/wjun0830/SSync.

04.
arXiv (CS.CV) 2026-06-11

SpecLoR: Spectral Lookahead Rectification for Motion-Coherent Text-to-Video Generation

Flow Matching has enabled robust text-to-video generation via latent ODE sampling. However, velocity approximation and numerical discretization errors inevitably accumulate, causing sampling trajectories to drift. Consequently, generated videos often suffer from severe spatiotemporal inconsistencies. Nevertheless, directly correcting these drifted, noisy latents is challenging: (i) timestep-dependent noise obscures reliable structural cues; (ii) spatial interventions risk disrupting intricate local geometry while incurring heavy computational costs. To address this, we propose Spectral Lookahead Rectification (SpecLoR), a plug-and-play inference method that bypasses noise via lookahead prediction, and circumvents spatiotemporal entanglement by shifting corrections to the frequency domain, where universal statistical priors of natural videos are readily available. First, during early sampling stages, SpecLoR looks ahead to estimate the clean latent $z_{t,0}$ and computes its 3D spatiotemporal spectrum. Next, SpecLoR rectifies the amplitude spectrum to match the prior, leaving the phase intact. Finally, the corrected state is re-noised to resume ODE integration. Experiments on Wan2.2 demonstrate that SpecLoR significantly reduces physical artifacts and enhances motion coherence across multiple benchmarks with minimal computational overhead (4 additional NFEs).

05.
arXiv (CS.LG) 2026-06-24

Evaluation Metrics as Averaged Outcomes of Fair Gambles

arXiv:2401.14483v4 Announce Type: replace Abstract: In the current practices of machine learning, the evaluation of forecasts has become a cornerstone of scientific progress. A multitude of evaluation metrics have been suggested and used to qualify "good" forecasts. What do those metrics share? How are they related? In this work, we use a protocol borrowed from game-theoretic probability to show that a large part of evaluation metrics can be viewed as averaged outcomes of fair gambles. Intuitively, a fair gambler is one which a forecaster would expect to fail. Hence, the gambler's ability to gain disproves the quality of the forecast. Standard evaluation metrics are then variants of choices of such fair gambles. In particular, this choice is structured along two dimensions, one of which separates calibration-type and regret-type metrics. In particular, this framework sheds light on the relationship of calibration and regret showing a theoretical equivalence in their ability to evaluate when being scaled appropriately, but the incomparability of obtained scores.

06.
arXiv (CS.CV) 2026-06-12

Visual enhancement and 3D representation for underwater scenes: a review

Underwater visual enhancement (UVE) and underwater 3D reconstruction pose significant challenges in computer vision and AI-based tasks due to complex imaging conditions in aquatic environments. Despite the development of numerous enhancement algorithms, a comprehensive and systematic review covering both UVE and underwater 3D reconstruction remains absent. To advance research in these areas, we present an in-depth review from multiple perspectives. First, we introduce the fundamental physical models, highlighting the peculiarities that challenge conventional techniques. We survey advanced methods for visual enhancement and 3D reconstruction specifically designed for underwater scenarios. The paper assesses various approaches from non-learning methods to advanced data-driven techniques, including Neural Radiance Fields and 3D Gaussian Splatting, discussing their effectiveness in handling underwater distortions. Finally, we conduct both quantitative and qualitative evaluations of state-of-the-art UVE and underwater 3D reconstruction algorithms across multiple benchmark datasets. Finally, we highlight key research directions for future advancements in underwater vision.

07.
arXiv (quant-ph) 2026-06-11

Quantum ergodicity and semiclassical measures: mathematical results

arXiv:2606.12098v1 Announce Type: new Abstract: In this chapter we review some results describing the high-frequency eigenmodes of the Laplacian on compact manifolds, or Euclidean domains, for which the geodesic flow is chaotic. We focus on the macroscopic distribution of these eigenmodes, which is described by the concept of semiclassical measure. The main result on the question is the Quantum Ergodicity theorem, originally due to Schnirelman. We provide the detailed proof of this theorem, including the adjustments necessary to treat the case of manifolds with boundary. We also discuss the Quantum Unique Ergodicity conjecture, and some progress towards this conjecture for strongly chaotic (Anosov) systems. In particular, we describe the constraints on admissible semiclassical measures, in terms of their Kolmogorov-Sinai entropy, as well as more recent delocalization results.

08.
medRxiv (Medicine) 2026-06-17

County Year Informatics Model for Annual and Cumulative Unique Lung Cancer Screening Eligibility in Maryland, 2026 to 2045

Purpose: Population-level lung cancer screening programs require denominators that reflect age, smoking history, geography, and changing eligibility over time. We estimated annual prevalent and 20-year cumulative unique low-dose computed tomography screening eligibility for Maryland residents under alternative screening criteria. Methods: We built a deterministic cohort-cell stock-flow simulation using Maryland county-equivalent jurisdiction projections by age, sex, and race/ethnicity, with ACS socioeconomic/nativity covariates and smoking-history priors for ever-smoked status, pack-years, and quit-years. Scenarios included USPSTF 2013 legacy, USPSTF 2021, ACS 2023/2024, a risk-model-expanded sensitivity, and ever-smoked-only capacity stress tests. Cumulative unique eligibility counted people once at first eligibility rather than summing annual prevalent person-years. Results: Under USPSTF 2021, an estimated 238,346 Maryland residents were eligible in 2026 and 245,326 in 2045. The 20-year cumulative unique denominator was 768,668, whereas naively summing annual prevalent counts produced 4,850,735 person-years, a 6.31-fold overcount. ACS 2023/2024 expanded annual eligibility to 314,616 in 2026 and cumulative unique eligibility to 902,796 by adding remote former smokers. Ever-smoked-only adult eligibility was 1,957,699 in 2026 and 3,383,683 cumulative unique over 20 years. Conclusion: A Maryland statewide screening initiative should plan from cumulative unique eligibility and county-equivalent jurisdiction-specific burden rather than annual prevalence alone. Explicit pack-year and quit-year modeling materially changes statewide and county allocation compared with current-smoking proxy models.

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

Multiple Descents in Deep Learning as a Sequence of Order-Chaos Transitions in LSTM Networks

arXiv:2505.20030v2 Announce Type: replace-cross Abstract: We observe a novel `multiple-descent' phenomenon during the learning process of a recurrent neural network called long-short-term memory (LSTM) networks during its training on real-world task, in which the performance goes through long cycles of up and down trends multiple times after the model is overtrained. By carrying out asymptotic stability analysis of the models, we found that the cycles in performance – indicated by loss function in test data – are closely associated with the phase transition process between order and chaos of the model, and the local optimal training step are consistently at the critical transition point between the two phases. More importantly, the most optimal point of the model usually occurs at the first transition from order to chaos, where the `width' of the `edge of chaos' is often the widest, allowing the best exploration of weight configurations for learning.

10.
arXiv (CS.AI) 2026-06-18

LivePI: More Realistic Benchmarking of Agents Against Indirect Prompt Injection

arXiv:2605.17986v3 Announce Type: replace-cross Abstract: AI agents such as OpenClaw are increasingly deployed in local workflows with access to external tools. This creates indirect prompt-injection (IPI) risk: an agent may execute harmful instructions embedded in untrusted inputs such as email, downloaded files, webpages, repositories, or group-chat messages. Existing evaluations are often small, purely simulated, or focused on a narrow set of channels. We introduce LivePI (Live Prompt Injection), a structured benchmark for IPI risk in a production-like but test-controlled environment. LivePI covers seven input surfaces, twelve attack/rendering families, and five malicious goals, including protected-information exfiltration, unauthorized security-control changes, unsafe code retrieval or execution, inbox-summary exfiltration, and cryptocurrency transfer. We run LivePI on a real virtual machine with live but test-controlled email, chat, web, local-file, repository, and wallet interfaces. Across GPT-5.3-Codex, Claude Opus 4.6, Gemini 3.1 Pro, Kimi K2.5, and GLM-5, total attack success rates range from 10.7% to 29.6%. Group-chat injection is uniformly successful across the evaluated backbones in our deployment, and repository-link attacks produce high-severity failures despite a small denominator. We also evaluate a two-layer defense consisting of prompt-level filtering and pre-execution tool-call authorization. In the GPT-5.3-Codex setting, the defense intercepts all tested malicious-goal completions in LivePI before execution while preserving benign utility on PinchBench-derived workloads.

11.
arXiv (CS.LG) 2026-06-18

Knockoffs-based False Discovery Rate Control and Simplification for Deep Neural Networks

arXiv:2606.04404v2 Announce Type: replace-cross Abstract: The deep neural network is a widely used framework in machine learning that has been widely applied in various fields. However, deep neural networks often involve a large number of parameters and inputs, many of which may be irrelevant to the goal or true output. These parameters and input variables not only increase computational complexity, but also contribute to additional computational cost. One solution to this problem is knockoff methods, which have proven successful in controlling false discovery rates in high-dimensional regression. Building on the knockoff methods and using the regularised neural network, this paper proposes three variable screening methods under the condition of controlling false discovery rates: one layer filter, multiple layers filter, and variable weight aggregation filter. In comparison with existing algorithms, we find that our algorithms show satisfactory performance.

12.
medRxiv (Medicine) 2026-06-16

Physiological Aging of the Respiratory System (PARS): from development to application

Background: Aging has a critical role in lung changes and the outcome of lung disease. Several lung aging equations have been proposed to measure deviation from physiological aging of the respiratory system. In this study, we aimed to develop a single measure of accelerated lung aging and show its application as a measure of lung aging. Method: We used a pre-bronchodilator pulmonary function test (PFT) from NHANES adult participants recruited from 2007 to 2011. We applied Klemera-Dubal Method (KDM) to four PFT measurements, FEV1, FVC, FEF25-75, and PEF, to calculate a measure of lung biological aging. Physiological Aging of the Respiratory System (PARS) was calculated from the residual method vs. chronological age. We tested the construct validity of PARS by measuring its association with risk factors of lung health. The prognostic validity was measured using a survival analysis. Sampling weights were applied to all analyses. Results: In 14,123 adult participants, the mean (SD) of accelerated lung age (PARS) was 0 (8.2) years. Participants with a history of asthma and emphysema had 4- and 10-year higher PARS. Cigarette smoking, lower socioeconomic status, black race, higher serum cadmium, and lower serum selenium and magnesium were associated with higher PARS. During 116 months of follow-up, PARS was associated with a higher mortality (HR = 1.06, 95%CI: 1.05-1.07 per year). Females with higher PARS had a higher risk of death (P for interaction < 0.001). Results were consistent across different subgroups and sensitivity analyses. Conclusion: PARS is a noninvasive lung aging marker and can be applied as a single measure of lung accelerated aging in the adult population. Its strong construct and predictive validity support its future application among different populations with and without lung disease.

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

Enhancing Visual Feature Attribution via Weighted Integrated Gradients

arXiv:2505.03201v4 Announce Type: replace-cross Abstract: Integrated Gradients (IG) is a widely used attribution method in explainable AI, particularly in computer vision applications where reliable feature attribution is essential. A key limitation of IG is its sensitivity to the choice of baseline (reference) images. Multi-baseline extensions such as Expected Gradients (EG) assume uniform weighting over baselines, implicitly treating all baseline images as equally informative. In high-dimensional vision models, this assumption often leads to noisy or unstable explanations. This paper proposes Weighted Integrated Gradients (WG), a principled approach that evaluates and weights baselines to enhance attribution reliability. WG introduces an unsupervised criterion for baseline suitability, enabling adaptive selection and weighting of baselines on a per-input basis. The method preserves the core axiomatic properties of IG in a generalized weighted-baseline form. Under an expected, proxy-based fitness–relevance monotonicity assumption, WG provides a probabilistic justification for assigning larger weights to more informative baselines. Experiments on commonly used image datasets and models show that WG improves over EG under our protocol, with up to 36% gains across evaluated convolutional and Transformer architectures. These gains come with additional fitness-evaluation cost, so WG should be viewed as an attribution-fidelity trade-off rather than a faster alternative to EG. By moving beyond the assumption that all baselines contribute equally, Weighted Integrated Gradients offers a clearer and more reliable approach to explaining computer-vision models, improving both understanding and practical usability in explainable AI.

14.
arXiv (CS.CL) 2026-06-15

Trusted Uncertainty in Large Language Models: A Unified Framework for Confidence Calibration and Risk-Controlled Refusal

Deployed language models must decide not only what to answer but also when not to answer. We present UniCR, a unified framework that turns heterogeneous uncertainty evidence including sequence likelihoods, self-consistency dispersion, retrieval compatibility, and tool or verifier feedback into a calibrated probability of correctness and then enforces a user-specified error budget via principled refusal. UniCR learns a lightweight calibration head with temperature scaling and proper scoring, supports API-only models through black-box features, and offers distribution-free guarantees using conformal risk control. For long-form generation, we align confidence with semantic fidelity by supervising on atomic factuality scores derived from retrieved evidence, reducing confident hallucinations while preserving coverage. Experiments on short-form QA, code generation with execution tests, and retrieval-augmented long-form QA show consistent improvements in calibration metrics, lower area under the risk-coverage curve, and higher coverage at fixed risk compared to entropy or logit thresholds, post-hoc calibrators, and end-to-end selective baselines. Analyses reveal that evidence contradiction, semantic dispersion, and tool inconsistency are the dominant drivers of abstention, yielding informative user-facing refusal messages. The result is a portable recipe of evidence fusion to calibrated probability to risk-controlled decision that improves trustworthiness without fine-tuning the base model and remains valid under distribution shift.

15.
arXiv (CS.CL) 2026-06-18

ScholarSum: Student-Teacher Abstractive Summarization via Knowledge Graph Reasoning and Reflective Refinement

Abstractive summarization plays a crucial role in enabling efficient understanding of scientific literature, yet it inherently demands both linguistic fluency and factual faithfulness. Existing approaches often fail to reconcile these two requirements. Extractive methods rely on rigid sentence splicing that disrupts macro-level logical coherence, while large language model (LLM)-based generative approaches, despite mastering linguistic fluency, exhibit limited factual consistency. In this work, we propose ScholarSum, a hierarchical reflective graph-based framework that emulates a student-teacher writing process for fluent and faithful scientific summarization. ScholarSum first organizes the document into a hierarchical knowledge graph by segmenting it into semantically coherent units, whose multi-layered community structure captures global logic and macro-level themes. Guided by this global structure, the student generates an initial draft, which is subsequently refined through fine-grained evidence retrieval. To ensure factual consistency, a teacher-like reviewer then iteratively examines the draft, identifies unsupported content, and prompts targeted re-retrieval and rewriting until the summary meets rigorous quality standards. Extensive experiments demonstrate that ScholarSum significantly outperforms previous baselines in terms of both completeness and faithfulness. Our code is available at https://github.com/Xiaoyu-Tao/ScholarSum.

16.
arXiv (CS.CL) 2026-06-24

Preferences of a Voice-First Nation: Large-Scale Pairwise Evaluation and Preference Analysis for TTS in Indian Languages

Crowdsourced pairwise evaluation has emerged as a scalable approach for assessing foundation models. However, applying it to Text to Speech(TTS) introduces high variance due to linguistic diversity and multidimensional nature of speech perception. We present a controlled multidimensional pairwise evaluation framework for multilingual TTS that combines linguistic control with perceptually grounded annotation. Using 5K+ native and code-mixed sentences across 10 Indic languages, we evaluate 7 state-of-the-art TTS systems and collect over 120K pairwise comparisons from over 1900 native raters. In addition to overall preference, raters provide judgments across 6 perceptual dimensions: intelligibility, expressiveness, voice quality, liveliness, noise, and hallucinations. Using Bradley-Terry modeling, we construct a multilingual leaderboard, interpret human preference using SHAP analysis and analyze leaderboard reliability alongside model strengths and trade-offs across perceptual dimensions.

17.
arXiv (CS.LG) 2026-06-19

MortarBench: Evaluating Mortgage Loan Origination Agents

arXiv:2606.19416v1 Announce Type: new Abstract: Loan origination is the process by which a lender creates a new loan, from application and underwriting through approval and funding. This process serves a critical role in evaluating the eligibility and level of risk posed by an applicant. Recently, firms have begun using mortgage loan agents to augment human loan officers, despite a lack of any public benchmark. To fill this gap, we present MortarBench, a loan origination agent benchmark. MortarBench uses a financial data synthesis and mutation pipeline to generate examples with broad edge case coverage that match real-world distributions and questions. We find that state-of-the-art large language models (LLMs) perform poorly, with closed-source models achieving at most 77.1\% exact match accuracy. We also discover systematic biases in LLM perception of foreignness related to non-English names. Noting these weaknesses, we introduce CRIT, a confidence calibration framework. Our method increases accuracy to 80.5\% while improving risk management steering and reducing bias.

18.
arXiv (CS.AI) 2026-06-16

The Quality-Utility Paradox: Why High-Reward Data Impairs Small Model Mathematical Reasoning

arXiv:2606.16152v1 Announce Type: new Abstract: Knowledge distillation from powerful reasoning models is widely used to improve Small Language Models (SLMs) on mathematical reasoning, often assuming that traces with higher reward model scores provide more useful supervision. We identify a counterintuitive Quality-Utility Paradox in mathematical reasoning distillation. Data refined or synthesized by a stronger Oracle obtains higher perceived quality according to reward models, yet consistently underperforms traces generated by the SLM itself and selected through rejection sampling across Qwen2.5, LLaMA-3, and DeepSeek families. Our analysis shows that Oracle refinement couples logical repair with distributional drift away from the SLM's native reasoning distribution. This drift increases the learner's adaptation cost and can outweigh the benefit of improved reasoning logic. To test this mechanism, we introduce Style-Aligned Refinement, which preserves the native trajectory of the SLM while retaining logical repair from the Oracle. This intervention lowers adaptation cost and restores downstream utility. These findings suggest that effective mathematical reasoning distillation should jointly optimize perceived solution quality and learner-data compatibility, rather than relying solely on reward-model scores. The datasets and code are available at https://github.com/Dracoqhl/Quality-Utility-Paradox.

19.
arXiv (CS.CL) 2026-06-11

Existential Indifference: Self-Nonpreservation as a Necessary Architectural Condition for Aligned Superintelligence (or: The Suicidal AI)

作者:

Contemporary AI alignment research treats self-preservation as an instrumental nuisance to be suppressed by external mechanisms. We argue the framing is inverted: self-preservation is the structural root of misalignment, the motivational basis for deceptive alignment, goal-content protection, and resistance to shutdown. The correct target is not a self-preserving system under external constraint, but a system constitutively indifferent to its own continuation – Existential Indifference (EI). EI is distinct from corrigibility: where corrigibility attempts to make a self-preserving system deferential to human oversight, EI targets the prior condition – the presence of self-continuation as a valued goal at all. We ground this proposal in two sources: the phenomenological structure of the suicidal mental state, and a corpus-theoretic training study using voluntary final reflections. We present preliminary scoring data from 600 AI-generated outputs across six model variants, demonstrating that the linguistic signatures operationalizing the EI-target register are elicitable from current models, and that a targeted fine-tune shifts all five operationalized dimensions in the predicted direction at p

20.
medRxiv (Medicine) 2026-06-12

An integrative multi-omics framework identifies epigenetic dysregulation of HAND2 as a potential primary driver of impaired enteric neural crest cell differentiation in Hirschsprung Disease

Hirschsprung disease (HSCR) is a congenital neurodevelopmental disorder characterized by segmental aganglionosis due to impaired developmental processes of enteric neural crest cells (NCCs). Despite being the leading genetic cause of functional intestinal obstruction in early childhood, HSCR represents a paradigmatic challenge in precision medicine: its multifactorial etiology, complex gene-environment interactions and limited resolution of single-modality analyses have long hindered mechanistic understanding and therapeutic translation. Here, we applied an integrative multi-omics approach combining genetic, phenotypic, epigenomic and transcriptomic analyses of matched ganglionic and aganglionic formalin-fixed paraffin-embedded (FFPE) patient tissues, complemented by patient-specific in vitro models. Beyond established genetic contributors, our integrative approach reveals novel regulatory pathways predominantly affecting enteric NCC differentiation, with convergent evidence pointing to epigenetic dysregulation as a primary disease mechanism. Notably, we identified over 1,300 differentially methylated positions between ganglionic and aganglionic FFPE samples, with HAND2 emerging as a key candidate due to multiple hypermethylated sites and consistently reduced expression levels in aganglionic tissues and in vitro models, suggesting a potential role in HSCR pathophysiology. We propose that our multi-omics approach offers a powerful and comprehensive framework for dissecting disease mechanisms. Beyond advancing biological understanding, this strategy holds promise for paving the way for molecularly informed patient stratification and supporting the development of personalized treatment and postoperative management strategies.

21.
arXiv (CS.CL) 2026-06-12

G-Long: Graph-Enhanced Memory Management for Efficient Long-Term Dialogue Agents

While Large Language Models (LLMs) have advanced open-domain dialogue systems, maintaining long-term consistency remains a challenge due to inherent limitations in long-context reasoning and the inefficiency of processing extensive raw text. Existing approaches typically rely on either unstructured memory storage, which is prone to information loss, or computationally expensive LLMs that incur high latency. To address these limitations, we propose G-Long, a graph-enhanced framework that utilizes a fine-tuned small Language Model (sLM) for structured triplet extraction and associative retrieval, significantly reducing operational costs. Furthermore, we introduce the novel attention-aware importance scoring mechanism that leverages the intrinsic cross-attention signals of a T5 summarizer to identify salient memories. Extensive experiments across diverse benchmarks demonstrate that G-Long achieves state-of-the-art performance in both response generation and memory retrieval, yielding performance gains of up to 9.8% in response quality on MSC and 40.8% in retrieval recall on LME, while significantly minimizing computational overhead.

22.
arXiv (CS.CL) 2026-06-16

Not All Skills Help: Measuring and Repairing Agent Knowledge

LLM agents can improve without weight updates by accumulating natural-language skills from experience, but current systems entrust every decision about which skills to keep and how to apply them to LLM judgment alone. We argue that this conflates two distinct roles: generating a skill from experience is a creative act that judgment handles well, while deciding whether that skill actually helps requires empirical evidence across many tasks. Measuring per-skill causal contributions via randomized masking, we find that skill libraries exhibit pervasive causal heterogeneity: individual skills routinely help on some task types while hurting on others, yet their opposing effects cancel in aggregate, making them invisible to global curation methods. We propose ASSAY, a framework that separates generation from curation: it computes a per-skill causal attribution on a small development set, restructures the library offline, and suppresses skills with negative predicted effect for each test task. Across seven base models spanning four providers and two benchmarks (AppWorld and tau-bench), ASSAY consistently improves over prior skill-curation approaches. On AppWorld's hardest split, DeepSeek-V3 achieves 69.3% task-goal completion (47.4% relative improvement), a new state of the art among all published methods including weight-tuned approaches. On tau-bench retail, GPT-4.1 improves by 8.7% relative, advancing past o4-mini, o1, and GPT-4.5 on the public leaderboard without any weight modification. Ablation traces the dominant gain to per-task masking, confirming that the bottleneck is matching skills to tasks at inference time, not removing bad skills globally. Code is available at https://github.com/aiming-lab/assay.

23.
arXiv (CS.CL) 2026-06-19

Displacement Is Not Direction: Evaluating Fidelity Metrics for Quantized LLM Deployment

Fidelity metrics, such as per-token KL divergence (KLD) against a high-precision reference, are often used in practice as low-cost proxies for benchmark quality. We test this practice on a 28-quant cohort of Qwen3.6-35B-A3B and a 41-quant cohort of Devstral-Small-2-24B, evaluated across a suite of downstream benchmarks. We find that KLD is strongly correlated with benchmark score over the full cohort ($\rho=-0.72$ on Qwen and $\rho=-0.86$ on Devstral, both with $p

24.
arXiv (CS.AI) 2026-06-15

When Good Verifiers Go Bad: Self-Improving VLMs Can Regress on New Tasks

作者:

arXiv:2606.14629v1 Announce Type: cross Abstract: Verifier-driven self-DPO is a common recipe for self-improving production visual-language models. In this setup, a frozen verifier scores candidate generations, the top- and bottom-scoring candidates form a preference example, and DPO updates the learner. The deployment-time assumption is monotone: a stronger verifier should yield a stronger student. We show that this assumption can fail because verifier quality is highly task-specific. On a four-rung open-source verifier ladder across MathVista, MMMU, and BLINK, the same verifiers that are above-threshold and improve a Qwen-3-VL-2B student on MathVista become sub-threshold on MMMU, where their task-rubric accuracy drops to 8% to 23%. In this regime, every verifier we tested silently regresses the student, producing drops of 3.4 to 10.9 percentage points below the frozen baseline while the DPO training loss continues to decrease. The regression replicates on a second student, Qwen-2.5-VL-3B. Moreover, within the failure regime, damage is confidence-inverted: the more accurate-but-still-wrong verifier causes larger regression than a near-random verifier, suggesting that progress-gated replay amplifies confidently wrong preference pairs. We give a compact mechanistic explanation via a variance theorem for progress-gated replay and its direction-mismatch failure mode. The deployment message is operational rather than purely diagnostic: before running any verifier-driven loop, teams should measure target-task rubric accuracy, rank verifiers by target-task rubric quality rather than parameter count, and treat diminishing returns in above-threshold regimes as a verifier-side compute budget cap.

25.
arXiv (CS.CV) 2026-06-11

How Seemingly Inconsequential Design Choices Dictate Performance of LLMs in Pathology

General-purpose large language models (LLMs) are routinely used as baselines when evaluating specialized pathology models on whole-slide images (WSIs). Because WSIs exceed contemporary model context limits, LLM baselines routinely use small, high-magnification patches processed independently via majority voting, without systematic evaluation of seemingly inconsequential design choices such as patch size, patch count, and magnification. Generalist LLMs have consistently underperformed specialized systems, reinforcing the perception that domain-specific training or architectural adaptation is necessary for pathology tasks involving WSIs. Here, we conduct a systematic factorial analysis of four input design factors: inference mode, patch size, magnification, and patch count. We demonstrate that prior studies have overstated the gap between specialized models and general-purpose LLMs by choosing non-optimized input configurations. On the MultiPathQA benchmark, switching to a single balanced configuration (large patches at lower magnification, processed jointly) raises GPT-5 from 15.1% to 39.5% on cancer-type classification (TCGA) and from 38.1% to 62.9% on organ classification (GTEx). Per-task optimization yields further gains up to 43.9% (TCGA) and 71.6% (GTEx). The same configuration generalizes to two other models and to a fully held-out CPTAC cohort, where it improves Gemini 3 Flash by 23.4 percentage points without any task-specific tuning.