01.
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。
02.
Nature (Science)
2026-06-17
Towards Conversational AI for Disease Management
作者:
While large language models (LLMs) have shown promise in diagnostic dialogue1, their capabilities for effective management reasoning—including disease progression, therapeutic response, and safe medication prescription—remain under-explored. We advance the previously demonstrated diagnostic capabilities of the Articulate Medical Intelligence Explorer (AMIE)1−3 through a new LLM-based agentic system optimized for multi-visit clinical management and dialogue. To ground its reasoning in authoritative clinical knowledge, AMIE leverages Gemini’s long-context capabilities4, combining in-context retrieval with structured reasoning to align its output with up-to-date clinical practice guidelines and drug formularies. In a randomized, blinded virtual Objective Structured Clinical Examination (OSCE) study, AMIE was compared to 21 primary care physicians (PCPs) across 100 multi-visit case scenarios designed to reflect UK NICE Guidance and BMJ Best Practice guidelines. AMIE was non-inferior to PCPs in management reasoning as assessed by specialists and scored better in both preciseness of treatments and investigations, and in its alignment with and grounding in clinical guidelines. To benchmark medication reasoning, we developed RxQA, a multiple-choice question benchmark derived from two national drug formularies (US, UK) and validated by board-certified pharmacists. Though AMIE and PCPs both benefited from the ability to access external drug information, AMIE outperformed PCPs on higher difficulty questions. While further research would be needed before real-world translation, AMIE’s strong performance across evaluations marks a significant step towards conversational AI as a tool in disease management.
03.
arXiv (CS.LG)
2026-06-18
FOSC-X: An Extended Framework for Optimal Local Cuts and Non-Horizontal Cluster Selection from Clustering Hierarchies
作者:
arXiv:2606.18972v1 Announce Type: cross Abstract: Extracting a flat clustering solution from a hierarchy is a common task in practical cluster analysis and can be formulated as an optimisation problem. Existing approaches focus on finding a single optimal solution. We introduce FOSC-X, a framework for extracting the top-M globally optimal flat clusterings from local, non-horizontal cuts of a hierarchical cluster tree, while optionally enforcing constraints on the number of clusters. This enables automatic identification of multiple high-quality alternative clusterings that capture different aspects of the hierarchical structure. Without constraints, the top-M problem can be solved in polynomial time using dynamic programming, exploiting the property that locally optimal partial candidates within subtrees can be combined to form globally optimal solutions while automatically determining the number of clusters. However, this can lead to solutions with numbers of clusters that are ultimately undesirable – e.g., too large to be meaningful or practically analysed within a particular application domain. Imposing cluster-count constraints breaks the optimality property underlying the unconstrained dynamic programming approach, since locally optimal partial candidates may no longer combine into feasible globally optimal solutions. FOSC-X addresses this challenge through a dynamic programming strategy that maintains compact sets of feasible candidates using lower and upper feasibility bounds while pruning infeasible or dominated combinations. The resulting method guarantees optimal rankings of the top-M solutions with linear-time complexity in the number of cluster nodes and dataset size, both with and without cluster-count constraints. Experiments show that FOSC-X efficiently reveals alternative clustering structures overlooked by single-solution extraction methods.
04.
arXiv (CS.CV)
2026-06-19
3D-PLOT-LLM: Part-Level Object Tokens for 3D Large Language Models
作者:
3D multimodal large language models (3D MLLMs) describe a 3D object as a whole but cannot address, name, or reason about its parts. Prior part-aware attempts add segmentation decoders, heavier 3D encoders, or bounding-box grammars at substantial parameter cost. We take a fundamentally different path: we reorganize the input token stream so that parts become directly addressable through the LLM's own vocabulary. Our model, 3D-PLOT-LLM, partitions the frozen point encoder's patches into K locally coherent regions and inserts, before each region's patch tokens, a learnable per-region marker and a reserved vocabulary token ; a Marker-Space Refinement (MSR) module then conditions each marker on its region's spatial statistics and adjacency neighbors. The model thus cites parts in its output and follows prompts that refer to parts by token, a capability absent from prior object-level 3D MLLMs. To probe this interface, we construct PartVerse-QA, a vocabulary-level part-QA benchmark adapted from PartVerse mesh annotations (77K training pairs and 588 held-out queries on disjoint object splits), on which 3D-PLOT-LLM reaches caption-to-slots Jaccard 0.459 and Exact-match 13.78%, with a slot-to-caption GPT-4o judge of 44.68. On the 3DCoMPaT-GrIn part-aware grounded description benchmark, 3D-PLOT-LLM outperforms PointLLM, Kestrel, PARIS3D, and SegPoint on every text-output metric, and ShapeLLM on 3 of 4, with up to +3.03 GPT-4o judge over PointLLM. On Objaverse whole-object captioning, adding PartVerse-QA at Stage 2 yields +0.65 SBERT and +1.85 GPT-4o over PointLLM, and tops PointLLM-PiSA on 4 of 5 traditional metrics (SBERT, SimCSE, BLEU-1, METEOR) despite targeting a different (part-grounded) objective. All with under 1M new trainable parameters on a frozen point encoder, an order of magnitude below prior part-aware 3D MLLMs, and no segmentation decoder or bounding-box head.
05.
arXiv (quant-ph)
2026-06-17
Approximately Decoding the Colour Code
作者:
arXiv:2606.18035v1 Announce Type: new Abstract: Recently we showed that minimum weight decoding in the (6.6.6 planar) colour code is NP-hard. However, it remained an open question as to whether it was possible to approximate the minimum weight decoding arbitrarily closely in polynomial time. In this paper we prove that it is possible: for any $\varepsilon>0$ there is an polynomial time algorithm that, given a syndrome, can find an error-set generating that syndrome whose weight is at most $1+\varepsilon$ times the weight of the minimum weight decoding. As a consequence we see that, for any $\varepsilon>0$, there is a polynomial time algorithm that can correct all errors of weight up to $(1-\varepsilon)d/2$ in the distance $d$ colour code (so almost up to the theoretical $d/2$ limit). The polynomial we give is impractically large, but it does open the door for sensible polynomial time algorithms that approximate minimum weight decoding and, in particular, shows that approximate decoding is not NP-hard.
06.
arXiv (CS.AI)
2026-06-11
Erased but Not Forgotten: How Backdoors Compromise Concept Erasure
作者:
arXiv:2504.21072v3 Announce Type: replace-cross Abstract: The expansion of text-to-image diffusion models has raised concerns about harmful outputs, from fabricated depictions of public figures to sexually explicit imagery. To mitigate such risks, prior work has proposed concept erasure methods that aim to sever unwanted concepts from the model via fine-tuning, yet it remains unclear whether these approaches truly remove all links to the harmful concept or merely conceal superficial connections. In this work, we reveal a critical vulnerability, the Erasure Evasion Backdoor (EEB): an adversary binds a backdoor trigger to a concept slated for removal, and this malicious link survives subsequent erasure. We show that both black-box and white-box adversaries can instantiate this threat. Across six state-of-the-art erasure methods, including robust ones that explicitly search for alternative representations of the target concept, EEB consistently exposes harmful content: up to 82% success against celebrity-identity unlearning, up to 94% for object erasure, and up to 16 times amplification of explicit-content exposure. While EEB uncovers a blind spot in current erasure methods, it also provides a diagnostic tool for stress-testing future concept erasure techniques.
07.
arXiv (CS.CL)
2026-06-24
Rule2Text: A Framework for Generating and Evaluating Natural Language Explanations of Knowledge Graph Rules
作者:
Knowledge graphs (KGs) can be enhanced through rule mining; however, the resulting logical rules are often difficult for humans to interpret due to their inherent complexity and the idiosyncratic labeling conventions of individual KGs. This work presents Rule2Text, a comprehensive framework that leverages large language models (LLMs) to generate natural language explanations for mined logical rules, thereby improving KG accessibility and usability. We conduct extensive experiments using multiple datasets, including Freebase variants (FB-CVT-REV, FB+CVT-REV, and FB15k-237) as well as the ogbl-biokg dataset, with rules mined using AMIE 3.5.1. We systematically evaluate several LLMs across a comprehensive range of prompting strategies, including zero-shot, few-shot, variable type incorporation, and Chain-of-Thought reasoning. To systematically assess models' performance, we conduct a human evaluation of generated explanations on correctness and clarity. To address evaluation scalability, we develop and validate an LLM-as-a-judge framework that demonstrates strong agreement with human evaluators. Leveraging the best-performing model (Gemini 2.0 Flash), LLM judge, and human-in-the-loop feedback, we construct high-quality ground truth datasets, which we use to fine-tune the open-source Zephyr model. Our results demonstrate significant improvements in explanation quality after fine-tuning, with particularly strong gains in the domain-specific dataset. Additionally, we integrate a type inference module to support KGs lacking explicit type information. All code and data are publicly available at https://github.com/idirlab/KGRule2NL.
08.
arXiv (CS.CL)
2026-06-16
Learn-to-learn on Arbitrary Textual Conditioning: A Hypernetwork-Driven Meta-Gated LLM
作者:
Conventional LLMs may suffer from corpus heterogeneity and subtle condition changes. While finetuning can create the catastrophe forgetting issue, application of meta-learning on LLMs is also limited due to its complexity and scalability. In this paper, we activate the meta-signal of $\beta$ within the SwiGLU blocks, resulting in a meta-gating mechanism that adaptively adjusts the nonlinearity of FFN. A hypernetwork is employed which dynamically produces $\beta$ on textual conditions, providing meta-controllability on LLMs. By testing on different condition types such as task, domain, persona, and style, our method outperforms finetuning and meta-learning baselines, and can generalize reasonably on unseen tasks, condition types, or instructions. Our code can be found in https://github.com/AaronJi/MeGan.
09.
arXiv (CS.AI)
2026-06-16
SkillVetBench: LLM-as-Judge for Multi-Dimensional Security Risk Evaluation in Open-Source LLM Agent Skills
作者:
arXiv:2606.15899v1 Announce Type: cross Abstract: Open-source LLM agent ecosystems are growing rapidly, yet the security of community-contributed skills - modular tool definitions that extend agent capabilities - remains largely unvetted. The gap we fill: existing scanners operate at the code layer and are structurally blind to instruction-layer and multi-agent risk - natural-language directives that hijack an agent, exfiltrate data through encoded side channels, or chain harm across pipelines - so what is needed is a semantic, multi-dimensional vetting system rather than another signature matcher. We present SKILLVETBENCH, a live public leaderboard on Hugging Face that uses an LLM-as-Judge to vet agent skills. What is new: SARS (Skill Agentic Risk Score), a five-dimensional agentic-risk metric with a principled weighted formula for instruction-following systems. What is integrated: full CVSS v4.0 vector decomposition and a ClawHub dual-view that places our LLM-generated review beside the official marketplace verdict. What is demonstrated: drawing on our companion benchmark paper [ 1], the LLM-as-Judge stage achieves zero false negatives across 78 confirmed-malicious skills and zero false positives across 22 benign controls, while the best static baseline (SKILLSIEVE) still misses 15%; for instruction-layer categories such as Prompt Injection and Memory Poisoning, conventional tools miss between 89% and 100% of threats (e.g., CODEBERT detects none of nine memory-poisoning skills). Detection rates vary from 35% to 95% across four LLM evaluators, motivating ensemble scoring in production deployments.
10.
medRxiv (Medicine)
2026-06-15
Iron deficiency testing among people with incident heart failure in primary care
作者:
…
Background: Given around 50% of people with heart failure have a degree of iron deficiency, guidelines recommend screening. It is uncertain to what extent this is done in primary care and whether testing is equitable. Aim: To report the proportion of people with incident heart failure who undergo a ferritin test within 12 months. Design and setting: Retrospective primary care cohort study using Clinical Practice Research Datalink Aurum data, between 2016 and 2021. Methods: We report the proportion of adults with an incident diagnosis of heart failure who received a ferritin test within 12 months. Multivariable logistic regression was used to examine the odds of testing based on key demographic covariates and co-morbidities. Results: Among 105,749 individuals with an incident diagnosis of heart failure (mean age 71.6 years, SD 14.3), only 35,688 (33.7%) received a ferritin test within the subsequent year. Increasing age (odds ratio 1.25 per 10-year increase, 95% CI: 1.24-1.27), female sex (male sex OR 0.86, 0.84-0.89) and Asian ethnicity (OR 1.70, 1.59-1.80) were all associated with increased odds of testing as were diagnoses of coeliac disease (OR 1.86, 1.58-2.21), type 1 diabetes (OR 1.82, 1.51-2.19) and cirrhosis (OR 1.64, 1.43-1.87). There was geographic variation in testing, even in adjusted analyses. Conclusion: In a large primary care dataset, two thirds of people with incident heart failure did not receive a ferritin test for iron deficiency within a year of diagnosis demonstrating a gap in current practice and an opportunity for improvements in service delivery.
11.
arXiv (CS.CV)
2026-06-17
Plug-and-Adapt: Multimodal Coreference Resolution at First Sight with a Pretrained Alignment Model
作者:
Visual information helps resolve ambiguity in coreference resolution, leading to notable performance gains. However, existing Multi-modal Coreference Resolution (MCR) methods require training with (partially) annotated data from the target dataset before they can be applied, preventing their direct usability and raising concerns about generalization. While Vision-Language Large Models (VLLMs) with billions of parameters offer promising zero-shot capabilities, they remain largely inaccessible. Their massive size limits deployability, and many are only accessible through paid APIs. In this paper, we propose a plug-and-adapt method that strategically adapts a carefully pre-trained alignment model for immediate use in MCR tasks, designed to eliminate the need for training on scarce benchmark datasets or relying on resource-intensive VLLMs. Specifically, we first pre-train a fine-grained alignment model between textual and visual contextual information using vision-language alignment datasets. We then repurpose the alignment model to MCR through similarity aggregation by fusing visual and categorical cues with evidence theory, thereby enhancing effectiveness. Experiments on the Coreference Image Narratives (CIN) benchmark dataset demonstrate the effectiveness of our method, achieving a 5.31\% and 2.12\% improvement in CoNLL F1 over SOTA dedicated methods and popular VLLMs, respectively. We further evaluate our method on a masked CIN dataset for robustness testing and on a specially constructed VCR-MCR dataset for generalization assessment, with results confirming both capabilities.
12.
arXiv (CS.CL)
2026-06-15
Benchmarking Web Agent Safety under E-commerce Deceptive Interfaces
作者:
As autonomous web agents are increasingly deployed to perform real-world tasks, ensuring their safety has become a critical concern. In this work, we study web agent behavior under realistic deceptive interfaces in the e-commerce domain. We introduce WebDecept, a lightweight and configurable plugin framework that enables controlled injection of deceptive interface patterns into existing web environments. Using WebDecept, we instantiate seven deceptive patterns commonly observed on the open web, including targeted advertisements, domain redirection, and shopping manipulation. By injecting these patterns into the frontend during task execution, we perform controlled evaluation of multiple multimodal web agents. Our results show that current web agents are highly susceptible to multiple classes of deceptive interfaces, and that prompt-based constraints are often insufficient to mitigate these failures. We further analyze how the design choices of deceptive patterns influence the success of such manipulations. These findings highlight safety challenges that should be addressed as web agents are scaled toward real-world deployment.
13.
arXiv (CS.CL)
2026-06-16
Formalize Once, Edit the Rest: Efficient Lean-Based Answer Selection for Math Reasoning
作者:
With large language models (LLMs) increasingly applied to mathematical reasoning, formal proof assistants such as Lean can be leveraged to verify reasoning outputs with machine-checkable rigor, enabling use cases such as answer selection in test-time scaling with K sampled candidate answers. However, employing Lean requires that LLM outputs, originally in natural language, first be formalized. Existing Lean-based answer-selection work uses an autoformalization model to generate a formal statement in Lean for each candidate answer independently, incurring a significant computational cost. We propose BASE, a base-and-edit pipeline that formalizes a single base candidate per problem and derives the remaining K-1 statements by editing the answer expression in place. To facilitate this, we train a rewriter model LEANSCRIBE to localize the answer in the base formalization and generate a reusable edit function for the other K-1 candidates. BASE simultaneously improves selection accuracy and reduces formalization cost - a Pareto improvement that holds on all 12 (dataset, solver) configurations across four benchmarks and three solvers, cutting autoformalizer calls by about 5x at K=8, with the reduction expected to become larger as K grows. Code is available at https://github.com/ucr-rai/base-and-edit.
14.
arXiv (CS.LG)
2026-06-16
A Comparative Study of Graph Neural Network Layer Selection for Interaction Modelling in Driving Trajectory Prediction
作者:
arXiv:2606.14956v1 Announce Type: new Abstract: Autonomous driving systems rely on precise trajectory prediction to plan safe and efficient movement. Graph Neural Networks (GNNs) have become a promising approach for modelling spatiotemporal interactions among road agents. However, designing GNN architectures for trajectory prediction remains non-standardized, with little guidance on which graph layers effectively capture spatial interactions and temporal dynamics. This paper offers a detailed comparative study of 19 graph layer types, focusing on their spatial and temporal processing capabilities to discover the most effective architectures for trajectory prediction. Within the explored hyperparameter setting, we highlight five standout layer combinations, with ARMA, Chebyshev, and topology-aware layers consistently performing better than others. Beyond performance metrics, our findings yield practical design principles: sum-based aggregation is more effective than mean-based methods, multi-head attention mechanisms enable richer interactions, and assigning different weights to different hop distances significantly improves prediction accuracy. These findings offer useful guidance for designing more interpretable and effective trajectory prediction models.
15.
arXiv (CS.CV)
2026-06-18
Zero-Shot Cross-City Generalization in End-to-End Autonomous Driving: Self-Supervised versus Supervised Representations
作者:
End-to-end autonomous driving models are typically trained on multi-city datasets using supervised ImageNet-pretrained backbones, yet their ability to generalize to unseen cities remains largely unexamined. When training and evaluation data are geographically mixed, models may implicitly rely on city-specific cues, masking failure modes that would occur under real-world domain shifts when generalizing to new locations. In this work, we formulate zero-shot cross-city transfer as a controlled representation-level stress test for end-to-end autonomous driving and ask how visual pretraining affects transfer behavior under geographic domain shift. We conduct a comprehensive study by integrating self-supervised backbones I-JEPA, DINOv2, and MAE into planning frameworks. We evaluate performance under strict geographic splits on nuScenes in the open-loop setting and on NAVSIM in the closed-loop evaluation protocol. Our experiments reveal a substantial generalization gap when transferring models across cities with different road topologies, traffic conventions, and visual environments. In open-loop evaluation, a supervised backbone exhibits severe degradation when transferring between cities, yet some domain-specific self-supervised methods can substantially reduce both displacement and collision degradation. In closed-loop evaluation, self-supervised pretraining improves average out-of-distribution PDMS in several single-city training settings. Our results provide empirical evidence that representation learning influences the robustness of cross-city planning and motivate zero-shot geographic transfer as an important stress test for evaluating end-to-end autonomous driving systems.
16.
arXiv (quant-ph)
2026-06-19
Vine Codes: Low-Overhead Quantum LDPC Codes on a Planar Square Grid
作者:
arXiv:2606.20263v1 Announce Type: new Abstract: The surface code is a promising route towards large-scale quantum computing, requiring only nearest-neighbour gates amenable to superconducting hardware. However, surface codes incur large qubit overheads. Novel quantum low-density parity check (qLDPC) codes promise to reduce overheads but require long-range connections that are difficult to achieve on superconducting platforms. Here, we introduce "Vine Codes" - qLDPC codes that are implementable on a planar square grid through nearest-neighbour, two-qubit gates native to superconducting platforms (iSWAP and CZ). Our approach generalises "Directional Codes" recently introduced by Gehér et. al. (2025) which are constrained to a torus. In contrast, vine codes have open boundary conditions constructed with the aid of routing qubits. We perform extensive numeric searches and find promising candidate vine codes, e.g. [[121,4,6]], [[221,6,7]], and [[234,9,6]] codes. We verify the circuit distances and show that data and measure qubits required can be reduced by up to ~28% relative to the surface code at a circuit distance of 7. Even including routing qubits, vine codes require fewer total qubits than the surface code (e.g. ~18% reduction at circuit distance 10) and benefits are expected to increase at higher distances. We perform circuit-level noise simulations to demonstrate that under a realistic noise model and at a near-term noise rate of $10^{-3}$, vine codes can perform better than the surface code while using fewer qubits. We give an exhaustive list of all unique vine codes up to stabiliser-weight 9. We additionally introduce "Flip-Vine Codes" which possess single-qubit transversal Clifford gates useful for fault-tolerant logic and magic state cultivation. We furthermore construct examples of generalised open boundaries for vine codes that go beyond the familiar X/Z boundaries of the surface and tile codes.
17.
arXiv (CS.CV)
2026-06-24
Understanding Deep Representation Learning via Layerwise Feature Compression and Discrimination
作者:
Over the past decade, deep learning has proven to be a highly effective tool for learning meaningful features from raw data. However, it remains an open question how deep networks perform hierarchical feature learning across layers. In this work, we attempt to unveil this mystery by investigating the structures of intermediate features. Motivated by our empirical findings that linear layers mimic the roles of deep layers in nonlinear networks for feature learning, we explore how deep linear networks transform input data into output by investigating the output (i.e., features) of each layer after training in the context of multi-class classification problems. Toward this goal, we first define metrics to measure within-class compression and between-class discrimination of intermediate features, respectively. Through theoretical analysis of these two metrics, we show that the evolution of features follows a simple and quantitative pattern from shallow to deep layers when the input data is nearly orthogonal and the network weights are minimum-norm, balanced, and approximate low-rank: Each layer of the linear network progressively compresses within-class features at a geometric rate and discriminates between-class features at a linear rate with respect to the number of layers that data have passed through. To the best of our knowledge, this is the first quantitative characterization of feature evolution in hierarchical representations of deep linear networks. Empirically, our extensive experiments not only validate our theoretical results numerically but also reveal a similar pattern in deep nonlinear networks which aligns well with recent empirical studies. Moreover, we demonstrate the practical implications of our results in transfer learning. Our code is available at https://github.com/Heimine/PNC_DLN.
18.
arXiv (CS.LG)
2026-06-19
SSH-Net: A Deep Neural Network for Predicting Failure Time Distribution Functions under Competing Risks with Application to GPU Data
作者:
arXiv:2606.20451v1 Announce Type: cross Abstract: Competing risks are commonly observed in engineering fields and can bring challenges to time-to-event data modeling when the application scenarios are complicated. Recently, deep neural networks have received great attention for prediction with competing risks, due to their flexibility and high learning capability. However, the complexity of neural network structure brings extra difficulty in hyperparameter tuning based on different data inputs. Additionally, when an engineered system has complex physical structures with multiple hierarchical levels, treating all structural levels as a single group of inputs may fail to capture critical information. To address the issues, we propose a Structured Segmented Hazard Deep Neural Network (SSH-Net) for failure time prediction under cause-specific competing risks framework. Our approach associates neural network structure with data structures, and allows different covariate groups to impact the failure prediction through separate sub-networks. The neural network is constructed based on a cause-specific competing risks model. The SSH-Net outputs cause-specific hazard functions, and utilizes the penalized log-likelihood as the loss function. The prediction accuracy of SSH-Net is validated through simulation studies by evaluating the Brier score, the area under receiver operating characteristic curves (AUC), and the root mean square error (RMSE) of the predicted cause-specific cumulative incident function. We further demonstrate the model's ability to predict failure time distribution functions using the Titan GPU failure time data.
19.
arXiv (CS.CL)
2026-06-12
MoReBench: Evaluating Procedural and Pluralistic Moral Reasoning in Language Models, More than Outcomes
作者:
…
As AI systems progress, we rely more on them to make decisions with us and for us. To ensure that such decisions are aligned with human values, it is imperative for us to understand not only what decisions they make but also how they come to those decisions. Reasoning language models, which provide both final responses and (partially transparent) intermediate thinking traces, present a timely opportunity to study AI procedural reasoning. Unlike math and code problems which often have objectively correct answers, moral dilemmas are an excellent testbed for process-focused evaluation because they allow for multiple defensible conclusions. To do so, we present MoReBench: 1,000 moral scenarios, each paired with a set of rubric criteria that experts consider essential to include (or avoid) when reasoning about the scenarios. MoReBench contains over 23 thousand criteria including identifying moral considerations, weighing trade-offs, and giving actionable recommendations to cover cases on AI advising humans moral decisions as well as making moral decisions autonomously. Separately, we curate MoReBench-Theory: 150 examples to test whether AI can reason under five major frameworks in normative ethics. Our results show that scaling laws and existing benchmarks on math, code, and scientific reasoning tasks fail to predict models' abilities to perform moral reasoning. Models also show partiality towards specific moral frameworks (e.g., Benthamite Act Utilitarianism and Kantian Deontology), which might be side effects of popular training paradigms. Together, these benchmarks advance process-focused reasoning evaluation towards safer and more transparent AI.
20.
arXiv (CS.AI)
2026-06-16
CoffeeBench: Benchmarking Long-Horizon LLM Agents in Heterogeneous Multi-Agent Economies
作者:
arXiv:2606.16613v1 Announce Type: new Abstract: As LLM agents become capable of increasingly long-horizon tasks, evaluating their performance in economic systems is becoming increasingly important. Unlike existing benchmarks that primarily evaluate a single agent interacting with a passive environment, economic systems are inherently multi-agent, requiring autonomous agents to communicate, negotiate, and transact while pursuing their own objectives over extended periods. We introduce CoffeeBench, a benchmark for evaluating LLM agents in a long-horizon multi-agent economy composed of heterogeneous firms. In CoffeeBench, two farmers, two roasters, and two retailers autonomously operate their businesses over a 90-day simulation, each seeking to maximize cumulative net income through communication and transactions while managing cash, inventory, and pricing. The evaluated model controls one coffee roaster, while the remaining firms are controlled by fixed reference agents. Across several recent open-weight and proprietary LLMs, all models outperform a passive baseline that takes no actions, with most achieving positive net income. Analysis of agent behavior reveals substantial differences in long-horizon economic interaction: higher-performing models communicate more actively with other firms, whereas Claude~Haiku~4.5 exhibits an idle-drift failure mode, repeatedly choosing inaction despite producing coherent assessments and plans. We release our code and agent trajectories to support future research.
21.
arXiv (CS.CL)
2026-06-16
Why Tree-Style Branching Matters for Thought Advantage Estimation in GRPO
作者:
Group Relative Policy Optimization (GRPO) trains Chain-of-Thought reasoning with verifiable rewards, but estimating thought-level advantages without value functions often suffers from high variance. Although tree-style branching is used in practice to reduce variance, it lacks a theoretical explanation of why it works and whether it is important or potentially necessary. We study thought-level advantage estimation in GRPO from a variance perspective under a minimal tree-style setting where multiple continuations are sampled for each thought. Using the multivariate delta method, we reveal a sampling-dimension asymmetry. Increasing sampled thoughts ($K$) leaves a strictly positive estimation-variance floor, whereas increasing continuations per thought ($M$) drives the leading-order estimation variance to zero at rate $1/M$. This implies that, within the fixed-temperature GRPO-style estimator without value models studied here, accurate thought-level advantage estimation cannot be achieved by scaling thought sampling alone, making continuation-level branching a principled and potentially necessary mechanism rather than a heuristic. Experiments further provide empirical evidence for its effectiveness and potential necessity, demonstrating improved optimization stability, training efficiency, and final performance not only in math but also across vision domains and under different model architectures and sizes.
22.
arXiv (CS.AI)
2026-06-16
TimeVista: Exploring and Exploiting Vision-Language Models as Judges for Time Series Forecasting
作者:
arXiv:2606.16173v1 Announce Type: new Abstract: High-quality time series forecasting is pivotal for real-world decision-making. However, traditional point-wise metrics often fail to reveal complex temporal patterns and align poorly with human intuitive preferences. While the ''LLM-as-a-Judge'' paradigm has revolutionized text evaluation by providing flexible, human-aligned judgment, its application to time series remains largely unexplored. In this paper, we leverage Vision-Language Models (VLMs) as judges for time series forecasting, harnessing their ability to comprehend time series plots grounded in textual information. Specifically, we propose a novel framework integrating micro- and macro-level judgments informed by contextual information to evaluate time series forecasting. To this end, we introduce TimeVista, a comprehensive VLM-as-a-Judge benchmark comprising 5563 time series samples paired with detailed evaluation rubrics. Extensive meta-evaluations demonstrate that VLMs are highly reliable judges, achieving significantly higher consistency with human preferences than conventional metrics. Building upon our benchmark, we comprehensively assess recent Time Series Foundation Models (TSFMs) under the VLM-as-a-Judge paradigm. Our results demonstrate that VLMs serve as robust and interpretable judges, providing a comprehensive, human-aligned standard for evaluating time series models.
23.
arXiv (CS.AI)
2026-06-19
StreamKL: Fast and Memory-Efficient KL Divergence for Boosting Attention Distillation
作者:
arXiv:2606.20005v1 Announce Type: cross Abstract: Attention distillation, which trains one attention distribution to match another by minimizing their Kullback-Leibler (KL) divergence, is widely used in knowledge distillation, model compression, continual learning, and sparse-attention LLM training. However, existing approaches materialize both attention distributions before computing the KL reduction, incurring $O(N_QN_K)$ memory and IO costs that become prohibitive at long context lengths. We present StreamKL, the first fused GPU primitive for attention KL divergence that eliminates this quadratic materialization. StreamKL derives a novel online formulation for the coupled two-distribution KL reduction, enabling a single one-pass forward kernel that streams query-key tiles through on-chip SRAM. For the backward pass, StreamKL recomputes attention probabilities tile-by-tile, avoiding storage of quadratic intermediates. We further design and implement efficient GPU kernels with dedicated optimizations. Experiments show StreamKL delivers up to $43\times$ and $14\times$ speedups over baseline methods in the forward and backward passes, respectively. Most importantly, StreamKL reduces the extra HBM footprint of attention distillation from $O(N_QN_K)$ to $O(1)$, enabling long-context distillation on a single GPU.
24.
Nature (Science)
2026-06-17
Analysis of 173,303 exomes and genomes in the Pakistan Genome Resource
作者:
Naturally occurring loss-of-function variants in human genes enable drug target discovery because they mimic pharmacological inhibition of proteins. However, the study of these genetic variants is constrained by their rarity. Sequencing of diverse populations, particularly those enriched in familial relatedness, has been postulated to promote discovery of rare genetic variants1–3. Here we present the Pakistan Genome Resource, a South Asian biobank with high familial relatedness comprising 173,303 participants, who collectively carry naturally occurring homozygous loss-of-function variants in 6,476 genes. We describe the genetic architecture of this population, associations between genes and biomarkers, the distribution of loss-of-function variants across molecular pathways, and recall-by-genotype studies of therapeutically relevant genes. The Pakistan Genome Resource expands the catalogue of human genetic variants, provides a comprehensive genetic reference resource for the Pakistani population, and demonstrates the value of studying diverse cohorts to advance human health. The Pakistan Genome Resource compiles biobank data from 173,303 individuals with high familial relatedness, broadening the catalogue of human genetic variation and establishing a population-specific genomic reference for Pakistan.
25.
arXiv (CS.CV)
2026-06-16
MVM-IOD: An Industrial Object-Centric Benchmark Dataset for the Evaluation of 3D Reconstruction Methods
作者:
3D object reconstruction, and camera pose estimation in industrial applications are challenging tasks, as errors are costly while the computation time is often limited. The complexity of typical industrial objects further complicates these tasks. Most of the existing datasets in this context do not depict realistic industrial scenarios. Therefore, we introduce the Machine Vision Metrology Industrial Object Dataset (MVM-IOD). Images of typical industrial objects are captured systematically, by moving a camera, mounted at the end effector of an industrial robot arm, on a hemisphere around the objects. MVM-IOD contains reference camera poses and reference 3D point clouds, the acquired RGB images of 9 objects and 2 background choices resulting in 18 scenes, which allows evaluation of all image based methods that compute a 3D reconstruction, camera poses, or novel views of a scene. Based on MVM-IOD, we extensively evaluate current SOTA 3D reconstruction and camera pose estimation methods, such as Structure from Motion, Multi-View Stereo, recent feed forward methods (Visual Geometry Grounded Transformer, {\pi}3), and 2D Gaussian Splatting and report our findings as a baseline for future research. The experiments show that capture setups like ours generate out-of distribution images for feed forward methods, leading to suboptimal point clouds and camera poses. However, these out-of-distribution images can be shifted closer to the training distribution by applying simple preprocessing steps. Consequently, in certain industrial applications, feed forward methods should be used with caution.