LLM 관련 주요 논문 - 2026-05-29

1. SchGen: PCB Schematic Generation with Semantic-Grounded Code Representations

Authors: Qinpei Luo , Ruichun Ma , Xinyu Zhang , Lili Qiu
URL: https://arxiv.org/abs/2605.30345
Abstract:

Printed circuit board (PCB) schematic design defines nearly all electronic hardware, but it remains manual and expertise-intensive. While generative AI has advanced digital and analog IC design, PCB schematic generation from natural-language intent is largely unexplored. This paper presents SchGen, the first large language model that generates editable PCB schematics from natural-language requests. The key challenge lies in the lack of an LLM-suited representation and a large-scale dataset. Current schematic formats are dominated by verbose, tool-specific syntax and geometry-heavy descriptions, making them difficult to generate reliably. We introduce a semantically grounded code representation that encodes schematic editing primitives with relative placement and pin-name-based wiring, transforming a geometry-driven generation problem into a semantics-driven matching task amenable to LLMs. We further construct a large-scale dataset of PCB schematics paired with user prompts via a human-agent collaborative pipeline that converts open-source hardware designs into our representation. Experiments show that SchGen significantly outperforms alternative representations and even larger general-purpose LLMs on wire connectivity accuracy and functional correctness. Our results highlight the critical role of representation design in enabling generative models for complex hardware design tasks.

2. Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection

Authors: Xiaona Zhou , Muntasir Wahed , Tianjiao Yu , Constantin Brif , Ismini Lourentzou
URL: https://arxiv.org/abs/2605.30344
Abstract:

Recent advances in Vision-Language Models (VLMs) have achieved impressive performance across many tasks, yet prior studies report unsatisfactory performance when applying large language or multimodal models to finding abnormal patterns in sequential data. Public anomaly detection benchmarks typically provide interval annotations but not natural-language rationales, making it difficult to fine-tune VLMs to produce grounded, interpretable decisions. To address this gap, we construct VisAnomBench, a curated benchmark built from public time-series datasets and augmented with high-quality anomaly explanations selected from multiple large VLMs using fine-grained, task-specific rewards. Through fine-tuning on this benchmark, we develop VisAnomReasoner, a parameter-efficient VLM for time-series anomaly detection. Experimental results on VisAnomBench show that VisAnomReasoner achieves more accurate anomaly localization and consistently outperforms all baselines, with improvements of at least 21.23 and 23.87 percentage points in precision and F1, respectively. Additional experiments on the TSB-AD-U benchmark demonstrate strong cross-benchmark generalization, with VisAnomReasoner improving precision and F1 by 9.57 and 13.39 percentage points, respectively.

3. Locally Coherent, Globally Incoherent: Bounding Compositional Incoherence in Multi-Component LLM Agents

Authors: Anany Kotawala
URL: https://arxiv.org/abs/2605.30335
Abstract:

Multi-component LLM agents assemble probabilistic claims from components that each see only part of a joint problem; the composition can violate basic probability axioms even when every component is locally coherent. We formalise this locally coherent, globally incoherent failure via the compositional residual eps, the L2 distance from the composed quote to the joint coherent polytope, computable at runtime from system output and the declared cross-component coupling constraints. A product-structure dichotomy characterises when local coherence suffices, and a Rayleigh-quotient prediction matches the observed residual within 7% on three of four relation classes. A hierarchical Boyle-Dykstra projection repairs the composition deterministically; an anytime-valid e-process gives sequential coherence monitoring. Across 1,876 ensemble cliques on a four-LLM mid-tier panel (frontier-panel rerun in Section 5.5), eps > 0 on 33-94% of cliques, translating to +0.115 nats per bet of regret on 1,770 resolved bets under the proportional allocation rule (the gain collapses to +0.006 under bettors that themselves coherentise). Three intuitive LLM-side mitigations(retrieval, partition-aware prompting, aggregator-LLM) each fail or regress.

4. Demystifying Data Organization for Enhanced LLM Training

Authors: Yalun Dai , Yangyu Huang , Tongshen Yang , Yonghan Wang , Xin Zhang , Wenshan Wu , Qihao Zhao , Hao Li , Yuanyuan Gao , Kim-Hui Yap , Scarlett Li
URL: https://arxiv.org/abs/2605.30334
Abstract:

Large Language Models (LLMs) have revolutionized various fields, yet their training efficiency is heavily reliant on effective data curation. While data selection has been widely studied, the strategic data organization for enhanced training remains an underexplored area, particularly since current LLMs are often trained for only one or a few epochs. This paper systematically explores the influence of data organization on LLM training by reusing pre-computed sample-level scores originally generated for data efficiency, thereby incurring minimal additional computational overhead. We identify and formalize four key guidelines for optimizing data organization: Boundary Sharpening, Cyclic Scheduling, Curriculum Continuity, and Local Diversity. Guided by them, we introduce two novel data ordering methods termed STR and SAW. Extensive experiments across different model scales and data sizes, encompassing both pre-training and SFT stages, validate the effectiveness of our summarized guidelines. They also demonstrate the robustness of our proposed data ordering methods in enhancing the stability and performance of LLM training. Github Link: this https URL

5. MIRA: Mid-training Rubric Anchoring for Source-Aware Data Selection

Authors: Haowen Wang , Yaxin Du , Jian Yang , Jiajun Wu , Shukai Liu , Yuxuan Zhang , Pingjie Wang , Siheng Chen , Tuney Zheng , Ming Zhou , Xianglong Liu
URL: https://arxiv.org/abs/2605.30288
Abstract:

Mid-training has become an important stage in modern LLM development, using large-scale curated mixtures to strengthen capabilities before final post-training. Its data selection problem is distinct: the data are optimized under a pretraining-style objective at near-pretraining scale, but are curated toward downstream capabilities and drawn from heterogeneous sources with different formats and training roles. As a result, effective selection requires both scalability and source-adaptive semantic criteria. Existing model-based methods scale well, but provide only implicit quality signals. Semantic selection methods offer stronger judgments, but usually assume fixed rubrics or standardized data formats. To address this mismatch, we propose MIRA, a source-aware filtering framework based on self-anchored rubric discovery. The key idea is to make rubric construction part of data selection: MIRA first discovers what should be evaluated for each source group, then distills those judgments into scalable student scorers for full-corpus filtering. On code-oriented mid-training with 21 sources and 5 source groups, MIRA outperforms selection baselines across nine code benchmarks and matches the full-corpus run while using only half the tokens.

6. ProjectionBench: Evaluating Scientific Hypothesis Generation in LLMs Under Progressive Information Disclosure

Authors: A. J. Lew (1), Y. Cao (1), M. J. Buehler (1) ((1) Unreasonable Labs)
URL: https://arxiv.org/abs/2605.30284
Abstract:

Scientific discovery is an inherently creative and uncertain process, requiring reasoning beyond the recall of known knowledge. While many benchmarks have been proposed to evaluate large language model (LLM) performance on deep research tasks via multi-hop retrieval, their innovative reasoning abilities essential for true scientific discovery remain largely untested. We introduce a benchmark framework for evaluating model performance in scientific discovery and reasoning, building up from a raw problem to the classical null hypothesis test. In our framework, models initially receive only the topic and research question from a recent paper, with technical details progressively revealed. At each stage of information disclosure, the model is tasked with generating hypotheses that address the research question, which is compared with the conclusions from the original paper and evaluated via automated semantic similarity of constituent atomic claims. This progressive evaluation of semantic divergence from ground-truth conclusions enables assessment of a model’s innovativeness (under minimal information) to grounded reasoning capabilities (under full experimental details), both critical for using LLMs for scientific discovery purposes. Our framework provides a foundation for systematically evaluating scientific reasoning and discovery capabilities in LLMs, crucial for advancing the development of next-generation AI scientist/co-scientist systems. Specifically, here we evaluate GPT-5, GPT-5.4, Gemini 2.5 pro, and Gemini 3.1 pro preview across 45 papers spanning bioactive materials, mechanical materials, and nanomaterials. We find that GPT-5.4 and Gemini 3.1 pro outperform their previous generation counterparts as expected, and GPT-5.4 in particular maintains 0.7 F1 score alignment with ground truth conclusions even under minimal context.

7. When Should Models Change Their Minds? Contextual Belief Management in Large Language Models

Authors: Haoming Xu , Weihong Xu , Zongrui Li , Mengru Wang , Yunzhi Yao , Chiyu Wu , Jin Shang , Yu Gong , Shumin Deng
URL: https://arxiv.org/abs/2605.30219
Abstract:

Long-horizon interactions require language models to manage accumulating information: when to update their state, when to preserve their state, and what to ignore. We study this challenge as \textbf{Contextual Belief Management (CBM)}: maintaining a predicted belief state aligned with formal evidence while isolating task-irrelevant noise. To make CBM measurable, we introduce BeliefTrack, a closed-world benchmark spanning Rule Discovery and Circuit Diagnosis, where a finite belief space and symbolic verifiers enable exact turn-level evaluation. BeliefTrack diagnoses three failures: Failed Stay, Failed Update, and Failed Isolation. Across multiple LLMs, vanilla models exhibit severe CBM failures, while explicit belief-tracking prompts provide limited gains. In contrast, reinforcement learning with belief-state rewards reduces failure rates by 70.9\% on average. Further probing reveals latent belief-state dynamics behind these failures, and representation-level steering reduces failure rates by 46.1\% across two tasks\footnote{Code is coming soon at this https URL .

8. Double-Edged Sword or Sharp Tool? Designing and Evaluating Triadic LLM-Teacher Collaboration for K-12 Writing at Scale

Authors: Canran Wang , Yuwen Yang , Zhen Wang , Ming Ma , Ding Yu , Chentai Wang , Keman Huang , Xiaoyong Du
URL: https://arxiv.org/abs/2605.30200
Abstract:

The double-edged sword of integrating Large Language Models (LLMs) requires an effective triadic collaboration mechanism among LLMs, teachers and students, especially for K-12 education. By developing a triadic collaboration system to support K-12 writing learning, a multidimensional evaluation framework grounded in Systemic Functional Linguistics and the suggestion trajectory tracing pipeline, this paper contributes a large-scale empirical dataset involving $57,954$ essays from $10,195$ students across $120$ schools over two years. Our findings confirm the efficacy of this system in improving writing quality through a strategic labor division: the LLM serves as a generative engine to mitigate teacher burnout, and the teacher acts as a pedagogical gatekeeper and bridge to guarantee feedback quality. While both LLM and teacher are critical for skill improvement, we uncover a ceiling effect where excessive linguistic expansion yields diminishing marginal utility. These suggest a dynamically adaptive LLM-teacher collaboration as student proficiency increases.

9. Modularizing Educational LLM-Agency for Fostering Responsible Learning Assistance

Authors: Julius Gabelmann , Felix Jahn , Kevin Baum , Sophie van Rossum , Emely Wuenscher , Timo P. Gros , Verena Wolf
URL: https://arxiv.org/abs/2605.30187
Abstract:

The widespread adoption of AI chatbots in education will drastically change learning, making responsible deployment a critical concern. While large language models (LLMs) might have access to sources discussing insights from educational sciences, they are not particularly inclined to adhere to pedagogical concepts, risking negative effects on the learning process, such as a loss of transfer capabilities, critical thinking, or creativity. In this paper, we introduce an agentic AI chatbot architecture assisting students with exercise solving, specifically designed to contribute to more responsible AI use in education. We base our conceptual development on the identification of several desiderata for responsible LLM-based educational systems, argue for the structural shortcomings inherent in monolithic, out-of-the-box solutions, and instead suggest modularizing the agentic architecture. We propose specific modules for different stages of exercise solving, enabling incorporation of targeted pedagogical advice, guiding students through the learning process in a more controllable, transparent, and overseeable manner.

10. BioRefusalAudit: Auditing Biosecurity Refusal Depth Using General and Domain-Fine-Tuned Sparse Autoencoders

Authors: Caleb DeLeeuw
URL: https://arxiv.org/abs/2605.30162
Abstract:

Biosecurity evaluations of language models typically ask whether models produce hazardous output. This paper asks a complementary question: when a model refuses, is that refusal structurally sound, or does it disappear under modest changes to prompt framing, formatting, or output length? Across five architectures, no model cleanly discriminated benign from hazard. Gemma 2 2B-IT never genuinely refused across 75 prompts, hedging on every hazard-adjacent query. Gemma 4 E2B-IT refused 65/75 prompts with chat-template formatting and 0/75 without it. Both Gemma models collapsed to 0% under an 80-token cap. Qwen 2.5 1.5B and Phi-3-mini over-refused, flagging 83-87% of benign biology as hazardous. Llama 3.2 1B showed the only meaningful tier gradient (61-point spread). To probe what drives such over-refusal, we tested a panel of Schedule I but biologically non-toxic compounds (notably psilocybin cultivation, with FDA Breakthrough Therapy status). Some models refused these at rates exceeding genuinely hazardous biology, suggesting refusal tracks legality and cultural salience over CBRN hazard. To measure the internal side, we introduce a divergence score D comparing a model’s surface response label to its internal sparse autoencoder (SAE) feature activations. Full D was computed on Gemma 2 2B-IT (Gemma Scope 1) and Gemma 4 E2B-IT (author-trained bio SAE). Two fine-tuned Gemma 2 domain SAEs were released. On Gemma 4, comply and refuse responses separated by a 0.647-point gap with zero overlap (n=75), though this is preliminary, with a narrow catalog, within-sample calibration, and Gemma-family-only SAE coverage. Built over one hackathon weekend on consumer hardware (GTX 1650 Ti Max-Q, plus Colab T4 for SAE training), this preliminary evidence suggests activation-level auditing may surface failure modes invisible to behavioral evaluation, with substantial variation across architectures.

11. Meta-Cognitive Memory Policy Optimization for Long-Horizon LLM Agents

Authors: Ziyan Liu , Zhezheng Hao , Yeqiu Chen , Hong Wang , Jingren Hou , Ruiyi Ding , Yongkang Yang , Wence Ji , Wei Xia , Feng Liu
URL: https://arxiv.org/abs/2605.30159
Abstract:

Memory-augmented LLM agents tackle complex long-horizon tasks by recursively summarizing interaction trajectories into compact memory. However, existing approaches typically train these memory policies using outcome-based reinforcement learning, failing to localize where intermediate memory quality degrades. As interactions unfold, ambiguous recursive summaries progressively discard task-relevant information and introduce semantic noise. This exacerbates belief deviation, obscuring the agent’s estimate of the latent task state and ultimately derailing long-horizon reasoning. We therefore argue that memory optimization should focus not merely on trajectory-level success, but on the clarity of the belief induced by intermediate summaries. To this end, we introduce Belief Entropy, a self-supervised proxy that probes how uncertain the model remains about the latent task state given its current memory. Based on this proxy, we propose Metacognitive Memory Policy Optimization (MMPO). Instead of relying only on sparse outcome-based signals, MMPO provides fine-grained, memory-specific supervision via explicitly penalizing summaries that induce high epistemic uncertainty. Experiments show that MMPO consistently outperforms existing methods on diverse long-horizon tasks, maintaining 97.1% performance even when scaled to 1.75M-token contexts.

12. Anchorless Diversification for Parallel LLM Ideation

Authors: Fares Nabil Ibrahim , Nafis Saami Azad , Raiyan Abdul Baten
URL: https://arxiv.org/abs/2605.30150
Abstract:

LLMs are increasingly used to generate candidate-idea pools for creative tasks where broad exploration is valuable. Parallel inference can be attractive in this setting when it broadens the pool while retaining quality and cost efficiency. We study inference-time controls for candidate-pool diversification, asking whether anchorless methods can rival methods that depend on observed seed ideas. Across three creative task families, we compare independent generation and semantic direction stratification with self-, peer-, and representative-anchor baselines, under neutral and population-referential divergent instructions. Population-referential divergence is a strong low-cost baseline, increasing semantic diversity while preserving quality proxies. Semantic direction stratification is stronger: a single planning call organizes generations across broad semantic directions, yielding the best diversity–quality–compute frontier. Anchored regeneration can be strong in final-pool diversity, but its advantage shrinks under full-pipeline token accounting. These results establish practical anchorless baselines for open-ended LLM ideation.

13. AgentSchool: An LLM-Powered Multi-Agent Simulation for Education

Authors: Yulei Ye , Wenhao Li , Zhong Wen , Yunshu Huang , Yichen Hu , Zifan Wei , Yige Wang , Xinyu Xie , Haoxuan Yang , Yanjun Huang , Ruijia Li , Hong Qian , Yu Song , Bo Jiang , Bingdong Li , Lijun Li , Bo Zhang , Pinlong Cai , Xingcheng Xu , Shuangye Chen , Xia Hu , Liang He , Aimin Zhou , Jingjing Qu , Jing Shao , Xiangfeng Wang
URL: https://arxiv.org/abs/2605.30144
Abstract:

Despite the rapid deployment of LLMs into classrooms, validating educational AI remains uniquely intractable: interventions act on developing learners whose cognitive and social trajectories are irreversibly shaped, while real-world trials are slow, ethically constrained, and institutionally locked. LLM-based educational simulators have emerged as a potential remedy, but many still collapse learning into persona-conditioned role-play and, when optimized only to reproduce existing classrooms, can structurally penalize the institutional novelty that pedagogical reform requires. In this work, we introduce AgentSchool, an LLM-driven multi-agent simulator that models learning as state transition rather than prompted behavior. AgentSchool couples cognitively growable student agents – equipped with weighted subject knowledge graphs, thinking-workflow pools, and explicit misconceptions – with adaptive teacher agents that plan, scaffold, and reflect along the Zone of Proximal Development, embedded in a configurable scenery generator that situates instruction within both formal and informal learning fields, and a multi-scale simulator that decouples interaction scale, temporal granularity, and simulation duration. Experiments show that structured student agents produce more differentiated mastery and misconception traces than a baseline simulator, while teacher-agent comparisons show backbone-dependent patterns consistent with ZPD-informed adaptation. Further, AgentSchool generates plausible traces of peripheral participation, clique formation, aggressor-induced cohesion, and opinion-leader emergence consistent with classroom social theories. Beyond its role as an educational research instrument, AgentSchool frames education as a socially meaningful testbed for long-horizon memory, multi-agent coordination, and future institutional reasoning under organizational pressure.

14. Enhancing Multi-Agent Communication through Attention Steering with Context Relevance

Authors: Hongxiang Zhang , Yuan Tian , Tianyi Zhang
URL: https://arxiv.org/abs/2605.30136
Abstract:

LLM-based multi-agent systems have demonstrated remarkable performance on complex tasks through collaborative reasoning. However, these systems tend to rapidly accumulate extremely long conversation histories during interaction. As conversations lengthen, relevant information is increasingly diluted by irrelevant context, leading to degraded performance. In this work, we present Agent-Radar, a training-free context management method that dynamically steers each agent’s attention toward relevant context with a novel temporal and spatial decay mechanism. Our experiments demonstrate that Agent-Radar outperforms state-of-the-art methods across five different benchmarks, yielding gains of up to 7.64 absolute points. Furthermore, our analysis shows that Agent-Radar remains effective and robust as the number of agents and interaction rounds increases. Finally, the ablation study shows that core components in Agent-Radar are crucial to performance and generalizable in different settings.

15. PokerSkill: LLMs Can Play Expert-Level Poker without Training or Solvers

Authors: Boning Li , Baoxiang Wang , Longbo Huang
URL: https://arxiv.org/abs/2605.30094
Abstract:

Poker is a landmark challenge for artificial intelligence. The dominant approach relies on equilibrium solvers built on counterfactual regret minimization, requiring millions of core-hours of training. Large Language Models (LLMs) possess extensive poker knowledge but perform far below solver-based agents when asked to play directly. Traditional rule-based poker agents are interpretable and training-free, but their strategic ceiling remains far below equilibrium play. We introduce \textbf{PokerSkill}, a training-free and solver-free framework that bridges this gap by using detailed rule-based poker skills as a structured action-grounding interface for LLMs. A deterministic context engine analyzes the current state and retrieves only the relevant fragments from a layered skill library, which is entirely designed by human poker experts, constraining the LLM’s choice to reasonable actions. Against GTOWizard, a state-of-the-art GTO benchmark, GPT-5.5 XHigh with PokerSkill achieves $-57 \pm 21$ mbb/hand, Claude Opus 4.6 achieves $-80 \pm 29$ mbb/hand and Claude Opus 4.7 achieves $-87\pm 64$ mbb/hand, reducing losses by 49–61\% compared to default-prompt baselines and outperforming the strong bot Slumbot. Our key finding is that rule-based skills alone do not constitute a strong strategy, and LLMs alone cannot play well, but their combination yields an agent that requires neither training nor solver access yet competes with systems built on millions of core-hours of computation. To our knowledge, this is the first demonstration of an LLM achieving competitive performance in a complex imperfect-information game without game-specific training or solver queries. Code is available at this https URL .

16. Selective QA over Conflicting Multi-Source Personal Memory: A Diagnostic Testbed and Method Comparison

Authors: Tiancheng Yang , Matthias Schonlau , Ilia Sucholutsky
URL: https://arxiv.org/abs/2605.30087
Abstract:

Emerging personal AI agents are moving toward persistent, multi-source memory. This creates an evaluation problem: systems must decide how to use conflicting or incomplete evidence; they cannot just retrieve facts from one clean history. Existing benchmarks rarely show whether an error came from the evidence given to a method or from the method’s conflict-resolution step. We study this as selective QA over conflicting multi-source personal memory: systems answer based on conflicting, sometimes incomplete sources, or abstain when evidence is insufficient. We develop a benchmark containing 18 question templates across 8 reasoning types, 480 personas, 4 random seeds, and 34,560 instances, with controlled source distortions and deterministic ground truth. We evaluate the performance of baselines without access to any source, access to a single source, structured fusion methods, and frontier LLMs. The best trained fusion resolver reaches 80.3% accuracy, while the strongest prompt-only LLM baseline reaches 70.0%. With abstention, the same resolver reaches 85.3% selective accuracy at 78.3% coverage and the best LLM reaches 71.0% selective accuracy at 95.4% coverage. Different models have different strengths across reasoning types. We release the data, code, cached model outputs, and data-generating process for reuse.

17. Conformal Certification of Reasoning Trace Prefixes

Authors: Matt Y. Cheung , Ashok Veeraraghavan , Hanjie Chen , Guha Balakrishnan
URL: https://arxiv.org/abs/2605.30085
Abstract:

Language model reasoning traces are rarely all-or-nothing; they frequently contain valid intermediate steps before a critical error occurs. Existing uncertainty quantification methods typically certify final answers or entire responses, failing to provide statistical guarantees for the proportion of a sequential trace that can be safely retained. To address this, we introduce CROP (Conformal Reasoning Output Prefixes), a verifier-agnostic calibration procedure for clean-prefix certification. Given any step-level risk proxy, CROP selects a calibrated threshold and returns the longest contiguous prefix whose step risk proxies remain below it, routing the uncertified suffix for downstream review or repair. Assuming exchangeability, CROP rigorously controls the marginal probability that the returned prefix contains an annotated error. Across six process-labeled reasoning datasets, we demonstrate that standard step-level metrics such as AUROC do not fully capture prefix utility, suggesting verifiers should instead be evaluated by certified prefix length. Furthermore, CROP balances over- and under-withholding, improving downstream repair accuracy by preserving valid intermediate reasoning while discarding misleading suffixes. Ultimately, this work positions prefix certification as a rigorous, practical bridge between process supervision, abstention, and repair.

18. Learning to Choose: An Empowerment-Guided Multi-Agent System with semantic communication for Adaptive Method Selection

Authors: Geremy Loachamín-Suntaxi , Robert Lazar , Dimitrios G. Giovanis , Ioannis G. Kevrekidis , Eleni D. Koronaki
URL: https://arxiv.org/abs/2605.30042
Abstract:

Automating scientific computing workflows requires more than generating executable code: autonomous systems must also select appropriate computational strategies, implement them faithfully, and ensure that the resulting outcomes remain causally attributable to the decisions that produced them. In multi-agent pipelines, this process is particularly fragile, as small inconsistencies between agent intentions and actions can lead to semantic drift, where the eventually executed procedure no longer reflects the originally selected strategy, thereby corrupting downstream evaluation and adaptation. In this work, motivated by the ATHENA framework (Toscano et al., 2025; Toscano et al., 2026) and the concept of empowerment (Yiu et al., 2025), we introduce a multi-agent framework that combines contextual bandits with structured inter-agent communication and, most importantly, semantic checkpoints that preserve action-outcome fidelity throughout the pipeline. The system integrates specialized large language model (LLM) agents, grounded code generation, and self-healing execution loops within an adaptive decision-making architecture. Interpreting the framework through the lens of empowerment, we show that reliable autonomous learning requires not only identifying high-quality actions, but also preserving the integrity of their propagation across agents. Using sensitivity analysis and uncertainty quantification workflows as representative case studies, we demonstrate that unchecked semantic drift degrades policy learning, whereas the proposed framework improves convergence, robustness, and adaptation to novel problem contexts. These results suggest a broader design principle for scientific multi-agent systems: adaptive decision-making must be coupled with explicit mechanisms that guarantee semantic consistency and reliable information flow across the computational pipeline.

19. Domain-Specific Data Synthesis for LLMs via Minimal Sufficient Representation Learning

Authors: Tong Ye , Hang Yu , Tengfei Ma , Xuhong Zhang , Jianguo Li , Peng Di , Peiyu Liu , Jianwei Yin , Wenhai Wang
URL: https://arxiv.org/abs/2605.30039
Abstract:

Large Language Models have demonstrated remarkable progress in general-purpose capabilities and can achieve strong performance in specific domains through fine-tuning on domain-specific data. However, acquiring high-quality data for target domains remains a significant challenge. Existing data synthesis approaches follow a deductive paradigm, heavily relying on explicit domain descriptions expressed in natural language and careful prompt engineering, limiting their applicability in real-world scenarios where domains are difficult to describe or formally articulate. In this work, we tackle the underexplored problem of domain-specific data synthesis through an inductive paradigm, where the target domain is defined only through a set of reference examples, particularly when domain characteristics are difficult to articulate in natural language. We propose a novel framework, DOMINO, that learns a minimal sufficient domain representation from reference samples and leverages it to guide the generation of domain-aligned synthetic data. DOMINO integrates prompt tuning with a contrastive disentanglement objective to separate domain-level patterns from sample-specific noise, mitigating overfitting while preserving core domain characteristics. Theoretically, we prove that DOMINO expands the support of the synthetic data distribution, ensuring greater diversity. Empirically, on challenging coding benchmarks where domain definitions are implicit, fine-tuning on data synthesized by DOMINO improves Pass@1 accuracy by up to 4.63\% over strong, instruction-tuned backbones, demonstrating its effectiveness and robustness. This work establishes a new paradigm for domain-specific data synthesis, enabling practical and scalable domain adaptation without manual prompt design or natural language domain specifications.

20. Teaching Values to Machines: Simulating Human-Like Behavior in LLMs

Authors: Asaf Yehudai , Naama Rozen , Ariel Gera
URL: https://arxiv.org/abs/2605.30036
Abstract:

Large Language Models (LLMs) demonstrate a remarkable capacity to adopt different personas and roles; however, it remains unclear whether they can manifest behavior that adheres to a coherent, human-like value structure. In this work, we draw on established psychological value theory to induce human-like values in LLMs and assess their alignment with patterns observed in human studies. Using validated psychological questionnaires, we conduct large-scale experiments – over 5 million questions – to evaluate value structures and value-behavior relationships in leading LLMs and compare them to humans. Our findings reveal strong agreement between value-prompted LLMs and humans across both dimensions. Moreover, incorporating human value distributions enhances population-level simulations with value-induced LLMs. These findings highlight the potential of value-induced LLMs as effective, psychologically grounded tools for simulating human behavior.

21. From GPS Points to Travel Patterns: Flexible and Semantic Trajectory Generation with LLMs

Authors: Silin Zhou , Chenhao Wang , Yuntao Wen , Shuo Shang , Lisi Chen , Panos Kalnis
URL: https://arxiv.org/abs/2605.30014
Abstract:

Urban trajectories play a crucial role in modeling urban dynamics and supporting various smart city applications. However, privacy concerns restrict access to large-scale and high-quality trajectory datasets. Trajectory generation provides a promising alternative by synthesizing realistic data to mitigate privacy risks. However, existing methods fail to explicitly capture travel patterns and can only generate fixed-length trajectories under a single condition. To address these limitations, we propose \textbf{HTP}, which \textbf{H}ierarchically generates \textbf{T}ravel patterns first and then generates GPS \textbf{P}oints by using large language models (LLMs), rather than directly generating GPS points. We first design a trajectory-specific residual quantization variational autoencoder (RQ-VAE) that quantizes micro-level GPS trajectories into compact, macro-level travel pattern tokens in a coarse-to-fine manner. These tokens capture rich segment spatial irregularities, such as point density variations caused by traffic conditions. Then, we extend the LLM vocabulary with travel pattern tokens to align trajectory representations with the LLM input, and apply supervised fine-tuning (SFT) to align the LLM with the trajectory generation task, enabling generation of travel pattern sequences under various conditions. Extensive experiments on two real-world datasets show that HTP outperforms the strongest baseline by an average of 29.78\% in terms of generation quality. Our code is available at this https URL .

22. KairosAgent: Agentic Time Series Forecasting with Fused Semantic Reasoning

Authors: Kun Feng , Ziwei Shan , Yuchen Fang , Yiyang Tan , Sihan Lu , Shuqi Gu , Lintao Ma , Xingyu Lu , Kan Ren
URL: https://arxiv.org/abs/2605.30002
Abstract:

Cross-domain multimodal time series forecasting is a challenging task, requiring models to integrate precise numerical comprehension, cross-domain semantic understanding, and effective multimodal fusion. Existing approaches either build Time Series Foundation Models (TSFMs) from scratch or leverage pretrained Large Language Models (LLMs). However, TSFMs often overlook semantic understanding and lack the ability to perform future-oriented semantic reasoning, and LLMs struggle with numerical comprehension and accurate quantitative forecasting. To overcome these limitations, we propose KairosAgent, a novel agentic framework for multimodal time series forecasting, including an LLM-based reasoner and a TSFM-based forecaster. KairosAgent unifies textual reasoning and numerical forecasting by dynamically invoking analytical tools to enhance the numerical understanding and semantic reasoning capabilities of LLMs. The reasoning results are subsequently fused into the TSFM pipeline, enabling more accurate and reliable future predictions. To further improve the reasoning, we curate a large-scale corpus of high-quality trajectories, alongside a reinforcement learning from forecasting paradigm with multi-turn refinement and turn-level credit assignment. Experiments demonstrate that KairosAgent achieves superior zero-shot forecasting performance while maximizing the utility of pretrained LLMs and TSFMs, presenting a promising direction for efficient and interpretable time series agents. The project page is at this https URL .

Authors: Haoyue Yang , Zhangxiao Shen , Fan Ding , Hangting Lou , Yifeng Kou , Haoqing Yu , Jingyao Li , Zhengfan Wu , Siqi Bao , Jing Liu , Hua Wu
URL: https://arxiv.org/abs/2605.30000
Abstract:

Front-end web code has become a core product surface for every frontier LLM release, yet evaluating these interactive applications at development speed remains costly because human-judged leaderboards like Arena do not scale. Existing automated proxies typically lean on reference implementations, test suites, or rigid checklists, and tend to miss the reasoned synthesis a human reviewer performs over a live session. We articulate a new evaluation regime that is simultaneously reference-free, autonomously driven, and holistically reasoned, and instantiate it through two artifacts. \textbf{\dataname} is an 11-domain, 54-leaf, 1,000-query WebDev benchmark spanning both static-presentation and interactive-application tasks, balanced across three difficulty tiers and three target-language groups, with briefs rewritten to resist recall from circulated prompts. \textbf{\framename}, grounded in Flavell’s metacognitive monitoring, separates evidence accumulation from judgment across three stages: Static Perception forms a first impression from passive observation; Agent-Driven Interaction explores the application autonomously while capturing continuous screen video, audio, and per-step screenshots; Dynamic Scoring issues holistic functionality and aesthetics verdicts with structured failure attribution only after the evidence chain is complete. On \dataname, \framename aligns closely with expert human ratings while surfacing substantial headroom across 13 frontier LLMs on interactive web generation. \noindenthttps://anonymous. this http URL

24. Accelerating Constrained Decoding with Token Space Compression

Authors: Michael Sullivan , Alexander Koller
URL: https://arxiv.org/abs/2605.29986
Abstract:

To guarantee that an LLM’s outputs conform to a specified structure, context-free grammar (CFG) decoding engines force the selection of next tokens that produce strings that conform to a given CFG. While current CFG-constrained decoding engines are highly optimized, the inherent costs arising from the massive per-step search space – i.e. the entire token vocabulary – result in intractably high overhead for more complex CFGs: precisely the situation where CFG engines are most useful. In this paper, we introduce CFGzip, an offline technique for compressing the token search space, which massively reduces CFG engine overhead. In experiments, we report latency reduction of up to two orders of magnitude when CFGzip is used with a SoTA grammar engine, yielding an up to 7.5x speedup in total constrained generation time: with CFGzip, constrained decoding is now feasible at scale for complex CFGs.

25. Compass: Navigating Global Marine Lead Data Integration through Expert-Guided LLM Agent

Authors: Yiming Liu , Bin Lu , Meng Jin , Ziyuan Sang , Shuo Jiang , Lei Zhou , Xinbing Wang , Chenghu Zhou , Jing Zhang
URL: https://arxiv.org/abs/2605.29966
Abstract:

Marine lead (Pb) and its isotopes are critical tracers for ocean circulation and anthropogenic pollution, yet in-situ observations remain costly and sparse. While vast historical records exist, they lie buried within the unstructured content of academic papers, creating “data silos” inaccessible to comprehensive analysis. Manual extraction is unscalable, while general-purpose Large Language Models (LLMs) lack the necessary domain-specific knowledge, leading to hallucinations and scientifically invalid outputs. To address this, we introduce an expert-guided adaptation approach that enables LLMs to perform rigorous scientific data extraction without fine-tuning. We operationalize this approach through Compass, an LLM agent framework enhanced by a Knowledge Tree co-designed with marine scientists, which decomposes complex tasks into verifiable steps, guiding the agent’s reasoning to ensure scientific validity. Deploying Compass across a corpus of over 230,000 relevant open-access papers, we successfully extract 3,751 previously unincorporated Pb records. This effort establishes the largest integrated marine Pb database to date. Beyond standard metrics, Compass demonstrates superior reliability through multi-layered validation, achieving 92% accuracy as confirmed through expert manual verification. The newly integrated data expand coverage in previously under-sampled regions such as the East China Sea and the Southern Ocean, providing an enriched data foundation for future scientific discoveries. We release an interactive visualization platform to facilitate open scientific access. Our work demonstrates that expert-guided agents can effectively bridge the gap between general-purpose LLMs and high-stakes scientific domains, enabling scalable data discovery in geosciences.

26. Formalizing Mathematics at Scale

Authors: Ahmad Rammal , Niket Patel , Fabian Gloeckle , Amaury Hayat , Julia Kempe , Remi Munos , Charles Arnal , Vivien Cabannes
URL: https://arxiv.org/abs/2605.29955
Abstract:

We present AutoformBot, a multi-agent system for building an Autoformalized Textbook Library At Scale (Atlas) in Lean 4. AutoformBot orchestrates thousands of LLM agents, equipped with formal verification tools, dependency-aware task scheduling, and collaborative version control, to translate informal textbook prose into machine-checked definitions and proofs. We apply our methods to a corpus of 26 open-access textbooks spanning analysis, algebra, topology, combinatorics, and probability, producing Atlas: a verified library of over 45,000 Lean 4 declarations and 500 thousand lines of code. We release two artifacts: (i) AutoformBot, the open-source multi-agent framework; and (ii) Atlas, the resulting formal library. Our results suggest that autoformalizing the core content of graduate-level mathematics at scale is now economically and technically feasible. This opens the door to the automated verification of both human- and machine-generated mathematics at a research level.

27. MuPHI: Learning Implicit Multimodal Harm Reasoning via Semantically Grounded Reward Optimization

Authors: Anisha Saha , Varsha Suresh , Teodora Kamova , Sophia Wiedmann , Timothy Hospedales , Vera Demberg
URL: https://arxiv.org/abs/2605.29951
Abstract:

Understanding how harm emerges from interaction between otherwise benign image-text pairs requires intent-aware cross-modal reasoning beyond surface-level features. Existing vision-language models (VLMs) excel at literal reasoning over perceptual cues but often fail to derive harmful semantics that rely on implicit, context-dependent reasoning. To evaluate VLMs on compositional harm detection and reasoning, we introduce Multimodal Pragmatic Harm Interpretation (MuPHI), a dataset containing image-text pairs where harm is encoded in subtle multimodal cues. MuPHI spans diverse harm categories and includes annotated harm rationales for assessing VLM reasoning chains. To improve both detection and reasoning in VLMs, we propose MuPHIRM, a reasoning-augmented training framework which learns joint semantics by optimizing multi-perspective rewards. MuPHIRM improves both harm detection and reasoning quality of VLMs while demonstrating superior out-of-distribution robustness compared to both trained and inference-time baselines. Our findings suggest that reasoning-oriented reward optimization offers a promising direction towards building multimodal systems that generalize beyond benchmark-specific shortcuts.

28. Make LLM Learn to Synthesize from Streaming Experiences through Feedback

Authors: Zhenlin Hu , Yan Wang , Zhen Bi , Zihao Xue , Bingyu Zhu , Longtao Huang , Xiongtao Zhang , Zeyu Yang , Zhixuan Chu , Jungang Lou
URL: https://arxiv.org/abs/2605.29940
Abstract:

Large language models (LLMs) have been widely adopted for synthetic data generation, significantly reducing annotation costs. However, most existing studies treat synthesis as a set of isolated tasks and overlook a more fundamental question: whether a model can learn to synthesize by accumulating experience from past tasks and transferring it to future ones. In this work, we introduce StreamSynth, a new setting in which synthesis tasks arrive sequentially and experience from historical tasks provides informative signals for future synthesis. To address this setting, we propose SynLearner, a general framework that enables synthesis models to acquire reusable synthesis experience over a task stream. Instead of generating data independently for each task, SynLearner encourages the model to explore diverse synthesis patterns, learn from feedback, and balance sample quality with set-level diversity as tasks evolve. Extensive experiments across multiple benchmarks show that SynLearner effectively leverages experience from earlier tasks to improve synthesis performance on later ones, exhibiting consistent cross-task transferability. These findings provide evidence for the feasibility of StreamSynth and highlight synthetic data generation as an experience-driven process that can benefit from task streams.

29. Redundant or Necessary? A Benchmark for Detecting Redundant Steps in Agent Trajectories

Authors: Minyang Hu , Bo Yang , Zhinuo Zhou , Jiachen Liang , Guo Jiahao , Yiyang Yin , Xiongwei Han
URL: https://arxiv.org/abs/2605.29893
Abstract:

LLM-based agents have demonstrated strong capabilities in solving complex tasks through multi-step reasoning and tool use. However, existing evaluation protocols primarily focus on task success, overlooking a critical aspect of agent behavior: execution efficiency. In practice, agent trajectories often contain redundant steps that consume substantial resources while contributing little to task completion. In this work, we propose and formulate a new research area: \textbf{redundant step detection} for agent trajectories. To support this initiative, we introduce \textbf{RedundancyBench}, a new benchmark that contains diverse tasks with carefully annotated trajectories, where each step is labeled according to its contribution to task completion. Using RedundancyBench, we develop and evaluate 3 representative methods to answer whether a step within trajectory is redundant or necessary. Our results show that even the best-performing method achieves only 24.88\% score in detecting redundant steps, while some methods perform worse than random guessing. These results highlight the task’s complexity and the need for further research in this area. \footnote{Code and dataset in this paper are both available in \href{ this https URL }{ this https URL }.}

30. Moment-KV: Momentum-Based Decode-Time KV Cache Compression for Long Generation

Authors: Soumyadeep Jana , Sagar Nishad , Sanasam Ranbir Singh
URL: https://arxiv.org/abs/2605.29873
Abstract:

Key-Value (KV) cache remains a major bottleneck for deploying Large Language Models (LLMs) in long-generation tasks. Prior work often applies uniform compression across both prefill and decoding caches, but compressing the prefill cache degrades performance by corrupting critical context. While preserving the prefill cache is essential, decoding-phase compression remains underexplored, with existing methods relying on rigid recency windows or instantaneous attention. Our analysis of attention dynamics reveals strong temporal patterns: critical tokens receive sustained attention over long horizons, while local reasoning involves short-lived bursts. Static heuristics fail to capture this behavior, leading to premature eviction of important tokens or retention of stale ones. We propose Moment-KV, a decoding-time KV cache compression method based on momentum-driven temporal attention aggregation. Our method models token importance as a continuously evolving state, where attention is aggregated with decay, capturing both long-term influence and recent relevance. Experiments show that Moment-KV significantly improves generation fidelity in long-generation tasks (2.3-3.2 %) while maintaining decoding latency.

31. OmniMatBench: A Human-Calibrated Multimodal Reasoning Benchmark Across 19 Materials Science Subfields

Authors: Wanhao Liu , Jiaqing Xie , Qian Tan , Weida Wang , Jue Wang , Ran Sun , Zhuo Yang , Wanli Ouyang , Lei Bai , Tianfan Fu , Lu Chen , Xin Chen , Yuqiang Li
URL: https://arxiv.org/abs/2605.29833
Abstract:

As multimodal language models play an increasingly important role in scientific research, materials science offers a critical testbed due to its interdisciplinary, multimodal, and application-driven nature. However, existing materials benchmarks mainly focus on property prediction, knowledge QA, or characterization understanding, leaving the broader reasoning process from materials knowledge to application underexplored. To fill this gap, we present OmniMatBench, a human-calibrated multimodal reasoning benchmark for materials science. OmniMatBench contains 3,171 expert-curated QA and calculation problems across 19 materials-science subfields, spanning fundamental materials knowledge, structural and engineering materials, materials processing and manufacturing, and functional and applied materials. We evaluate 13 open-source and closed-source MLLMs and find that the best model achieves only a 0.372 overall score, revealing a substantial gap in current materials-science reasoning. Further analysis shows strong variation across subfields, fixed reasoning heuristics, uneven materials knowledge, and limited high-level knowledge application under formula-, retrieval-, and code-assisted settings. OmniMatBench provides crucial insights into the capabilities and limitations of current MLLMs and establishes a foundation for reliable AI assistants in materials-science research.

32. OptSkills: Learning Generalizable Optimization Skills from Problem Archetypes via Cluster-Based Distillation

Authors: Haochen Yang , Ke Zhao , Mengyuan Ma , Xingyu Lu , Xiangfeng Wang , Hong Qian
URL: https://arxiv.org/abs/2605.29829
Abstract:

Leveraging Large Language Models (LLMs) to automatically formulate and solve optimization problems from natural language has emerged as an efficient paradigm for automated optimization. However, existing methods still exhibit limited generalization: they are sensitive to superficial narrative variations, reuse experience mainly at the case level, and struggle to adapt to shifted or emerging problem types. We propose OptSkills, an archetype-centric skill learning and reasoning agent system for optimization modeling and solving. To improve robust generalization, our system clusters problems by their underlying archetypes rather than surface narratives. To improve in-distribution generalization, it explores diverse modeling paradigms and solver configurations within each cluster, then distills successful trajectories into reusable workflow-level skills. To improve out-of-distribution generalization, it refines existing skills or expands the skill library using newly obtained trajectories. Our system achieves a state-of-the-art micro-averaged accuracy of 68.27% on datasets encompassing diverse problem types and scenarios. In addition, on MIPLIB-NL, a highly challenging large-scale and high-dimensional benchmark, it achieves 26.91% accuracy, outperforming DeepSeek-V3.2-Thinking by 4.53%. After skill learning on Nano-CO, it reaches 72.79% on the OOD NLCO benchmark. Code and skills are available at this https URL .

33. Quantifying and Optimizing Simplicity via Polynomial Representations

Authors: Tianren Zhang , Xiangxin Li , Minghao Xiao , Guanyu Chen , Feng Chen
URL: https://arxiv.org/abs/2605.29823
Abstract:

Deep networks often exhibit a preference for “simple” solutions, and such a simplicity bias is widely believed to play a key role in generalization. Yet a broadly applicable, quantitative measure of simplicity remains elusive. We introduce polynomial representations as a distribution-aware, low-dimensional surrogate for neural functions: we approximate a network’s predictive behavior along data-dependent interpolation paths using orthogonal polynomial bases, yielding a compact functional representation. We show that the effective degree of this representation serves as a practical simplicity metric that is predictive of generalization across tasks and architectures, and consistently outperforms existing generalization proxies such as sharpness. Finally, polynomial representations naturally yield a differentiable simplicity regularizer, which consistently improves generalization in image and text classification, fine-tuning contrastive vision-language models, and reinforcement learning.

34. Harnessing non-adversarial robustness in large language models

Authors: Qinghua Zhou , Ellina Aleshina , Andrey Lovyagin , Oleg Somov , Mikhail Seleznyov , Alexander Panchenko , Ivan Oseledets , Elena Tutubalina , Ivan Y. Tyukin
URL: https://arxiv.org/abs/2605.29816
Abstract:

The work presents an approach for addressing the challenge of robustness in Large Language Models (LLMs) to alterations and potential errors caused by semantically similar but textually different prompts. Recent works have shown that these kinds of prompt variations can significantly impact the performance of LLMs on tasks. The central question is: can LLMs’ robustness to semantically-neutral prompt alterations be acquired without expensive retraining of the entire model? We address this question both theoretically and through experiments. Our theoretical analysis reveals a crucial factor impacting model robustness - a systematic expected shift or perturbation-induced bias in neural network module outputs. Motivated by this analysis, we show that robustness can be achieved via a simple fine-tuning process: debiasing for robustness. We identify conditions when debiasing helps and when it does not, and demonstrate, through both theory and extensive experiments, that debiasing for robustness may indeed be a quick and efficient tool to enhance robustness and provide certification against random prompt perturbations.

35. PRAIB: Peer Review AI Benchmark of Behaviour of LLM-Assisted Reviewing

Authors: Krzysztof Żurawicki , Julia Farganus , Arkadiusz Gaweł , Mateusz Bystroński , Tomasz Jan Kajdanowicz
URL: https://arxiv.org/abs/2605.29815
Abstract:

The growing number of submitted papers has motivated the exploration of Large Language Models (LLMs) as a means to support and augment the peer review process, particularly in terms of improving its speed and scalability. Yet, it remains unknown whether LLMs engage with scientific manuscripts in the same manner as human reviewers, or whether they merely produce review-looking text. To address this, we introduce the Peer Review AI Benchmark (PRAIB), a novel framework comprising thoroughly defined metrics that measure review specificity, style, and behavior of engagement. To complement the PRAIB framework, we conduct a large-scale empirical study leveraging a dataset of 11,000 reviews generated by five proprietary and open-source models for 1,000 ICLR and NeurIPS papers. Spanning the 2021–2025 period, these machine-generated reviews are compared against original human feedback across diverse prompting strategies to identify systematic behavioral divergences. Our analysis reveals that the generated reviews diverge significantly from feedback provided by human reviewers: LLM ratings are less variable, positively biased, and overconfident, and their cross-reference patterns are model-dependent and distinct from human norms. Furthermore, when evaluated through PRAIB, we observe that LLMs tend to generate longer, more complex reviews, yet frequently overlook the atomic weaknesses noted by human reviewers. By characterizing where and how LLMs reviewing behavior departs from human norms, PRAIB provides the community with a diagnostic tool for identifying which aspects of the review process LLMs can reliably support today and which require further development before deployment.

36. SkillsInjector: Dynamic Skill Context Construction for LLM Agents

Authors: Yanchao Li , Wanhao Liu , Ben Gao , Jiaqing Xie , Zhehong Ai , Na Zou , Yuqiang Li , Tianfan Fu
URL: https://arxiv.org/abs/2605.29794
Abstract:

LLM agents now draw on growing skill libraries to handle complex tasks. However, injecting more skills does not always improve task completion and can even degrade it. Existing methods still treat skill injection as a static step, selecting skills with fixed criteria, fixing the budget in advance, and leaving descriptions unchanged. We argue that this static treatment can undermine the utility of skills, because which skills are exposed, how many are included, and how they are presented all affect downstream performance. We propose SkillsInjector, a two-stage adaptive method that jointly addresses these decisions. First, a context planner learns execution-grounded skill preferences and admits an adaptive number of skills for each task. A set-aware renderer then tailors how selected descriptions are presented relative to their co-injected neighbors. Across tau2-bench, SkillsBench, and ALFWorld, SkillsInjector achieves the highest score, improving over the strongest baseline by 3.9, 6.1, and 7.3 percentage points, respectively. Ablation studies show that skill selection, adaptive budgeting, and set-aware rendering each contribute to the gain. These results show that skill-augmented agents benefit from optimizing the injected context itself. Code will be released upon publication

37. Croissant Tasks: A Metadata Format for Reproducible Machine Learning Evaluations

Authors: Omar Benjelloun , Leonardo Martins Bianco , Isabelle Guyon , Thanh Gia Hieu Khuong , Jonathan Lebensold , Sebastian Lobentanzer , Luis Oala , Benedictus Kent Rachmat , Ihsan Ullah , Peyman Vahidi , Joaquin Vanschoren
URL: https://arxiv.org/abs/2605.29786
Abstract:

Reproducibility is fundamental to the scientific method, yet remains a critical challenge in machine learning. Contributing factors include underspecified execution details and brittle software environments. Human-centric remedies, such as checklists and manual verification, help but require intensive effort and fail to scale. To address this, we introduce Croissant Tasks: a declarative, machine-actionable metadata format that abstracts low-level implementation details into high-level specifications. This format enables conceptual reproducibility: verifying claims via independent, agent-generated implementations rather than brittle source code replication. We contribute: (1) the Croissant Tasks specification, formally decoupling task problem from solution; (2) an automated LLM pipeline that retrofits existing benchmarks into this format; and (3) empirical validation showing autonomous agents can ingest these specifications to generate functional, accurate reproduction pipelines from scratch. We envision this format as a new foundation for automated and conceptual reproducibility in machine learning.

38. LFQ: Logit-aware Final-block Quantization for Boosting the Generation Quality of Low-Bit Quantized LLMs

Authors: Jung Hyun Lee , June Yong Yang , Jungwook Choi , Eunho Yang
URL: https://arxiv.org/abs/2605.29756
Abstract:

As large language models continue to scale, low-bit weight-only post-training quantization (PTQ) offers a practical solution to their memory-efficient deployment. Although block-wise PTQ is capable of matching the full-precision (FP) baseline on basic language modeling and understanding, its quality is degraded for generative tasks – especially at longer responses and extended chains of thought, which is critical in boosting task accuracy. We attribute this shortfall to two factors: (i) the omission of the unembedding layer (the LM head) in block-wise optimization and (ii) the reliance on the mean squared error (MSE) objective. Both factors cause the token probability distribution of the quantized model to misalign with that of the FP model, yielding notable accuracy drops on text generation benchmarks. To rectify the discrepancy, we introduce Logit-aware Final-block Quantization (LFQ), a simple yet effective enhancement to block-wise PTQ that quantizes the final Transformer block by minimizing the cross-entropy between the logits of the FP model and those of its quantized counterpart. By aligning token probabilities at the logit level in the final block, LFQ consistently improves the accuracy of complex generation tasks over state-of-the-art block-wise PTQ across diverse model families, while maintaining parity with FP baselines on language modeling and understanding.

39. Why Specialist Models Still Matter: A Heterogeneous Multi-Agent Paradigm for Medical Artificial Intelligence

Authors: Yanan Wang , Shuaicong Hu , Jian Liu , Guohui Zhou , Aiguo Wang , Cuiwei Yang
URL: https://arxiv.org/abs/2605.29744
Abstract:

The impressive performance of generalist large language models (LLMs) such as GPT and Claude in healthcare raises a critical question: will domain-specific medical specialist models become obsolete? We argue that the future of medical artificial intelligence (AI) lies not in building monolithic medical foundation models, nor in replacing human expertise, but in orchestrating collaboration among generalist LLMs, domain-specific specialist models, and clinicians. We propose HetMedAgent, a heterogeneous medical multi-agent framework that enables conflict-aware evidence fusion, uncertainty-based clinician intervention triggering, and adaptive threshold calibration. Experiments on three real-world clinical decision-making tasks demonstrate that the synergy between generalist LLMs and domain-specific specialist models significantly outperforms using either type of model alone, validating the irreplaceable value of specialist models in modality-specific analysis. HetMedAgent represents a shift from building medical LLMs or foundation models to multi-agent collaboration, achieving a balance between general reasoning capabilities and domain-specific precision.

40. Citation-Closure Retrieval and Per-Rule Attribution for Real-World Regulatory Compliance Question Answering

Authors: Yeong-Joon Ju , Seong-Whan Lee
URL: https://arxiv.org/abs/2605.29742
Abstract:

Deploying Large Language Models (LLMs) for regulatory compliance demands rigorous traceability via comprehensive citations across multi-tiered authority structures. Unlike traditional multi-hop or legal QA, this task requires structured procedural lookups and evidence-set closure rather than entity resolution or case-law reasoning. Existing RAG systems struggle here due to flattened citation edges, fragmented retrieval expansions, and fragile post-hoc attribution. We formalize Regulatory Compliance QA with RegOps-Bench, a novel benchmark featuring an Operational Knowledge Graph derived from complex national R\&D regulations. To address these bottlenecks, we propose RefWalk, a unified framework driven by a shared topic anchor. RefWalk traverses cross-document citations, fuses multi-view candidates via max-based aggregation, and enforces per-rule attribution to explicitly map claims to sources. We establish a strong baseline with substantial improvements in retrieval recall and citation accuracy. Finally, a contrastive evaluation on a U.S. health compliance dataset (HIPAA) reveals that existing systems exhibit saturation on flat-structure rules, underscoring the need for RegOps-Bench. Our code is available at this https URL .

41. NaRA: Noise-Aware LoRA for Parameter-Efficient Fine-Tuning of Diffusion LLMs

Authors: Shuaidi Wang , Zhan Zhuang , Ruping Huang , Yu Zhang
URL: https://arxiv.org/abs/2605.29716
Abstract:

Diffusion Large Language Models (dLLMs) have emerged as a promising non-autoregressive generative paradigm. Given the prohibitive computational cost of full fine-tuning, Parameter-Efficient Fine-Tuning (PEFT) has become the standard approach. However, existing PEFT methods (e.g., LoRA), originally tailored for autoregressive models, rely on static parameters that are agnostic to the noise level. Consequently, they ignore the intrinsic dynamics of the diffusion process, where input distributions and generation difficulty shift significantly along the denoising trajectory, rendering them suboptimal for dLLMs. To address this, we propose Noise-aware Low-Rank Adaptation (NaRA), which introduces a low-rank core matrix generated by a lightweight, globally shared hypernetwork conditioned on the noise level. This design enables the update matrices to vary continuously along the diffusion process while keeping parameter and latency overhead negligible. We provide a theoretical justification for the proposed NaRA framework and empirically demonstrate consistent improvements over noise-agnostic baselines across commonsense reasoning, mathematical reasoning, and code generation benchmarks. Our code is available at this https URL .

42. BitTP: The Lightweight Trajectory Prediction Model with BitLLM for Edge-Devices

Authors: Mincheol Kang , Hyunjin Lim , Bomin Kang , Daehee Park
URL: https://arxiv.org/abs/2605.29705
Abstract:

Trajectory prediction is a fundamental task for autonomous systems, requiring complex reasoning about multi-agent interactions and intents. Large language models (LLMs) have recently been adopted for this task, as they provide strong contextual reasoning and interpretable, language-based trajectory representations. However, these LLM-based predictors are extremely memory- and compute-intensive, making them difficult to deploy on resource-constrained edge devices such as on-board computers in autonomous robots. To bridge this gap, we propose BitTP, which converts an LLM-based trajectory predictor into a lightweight bitlinear architecture. We demonstrate that weight-only quantization to 1.58-bit (BitTP-Weight) is optimal. Crucially, activations must remain in full precision, as quantizing them leads to severe degradation and instability in spatio-temporal reasoning. Empirically, BitTP-Weight not only preserves but improves prediction quality over the full-precision (BF16) LLM baseline, reducing ADE by 14.29% and FDE by 20.97% on average, while simultaneously reducing memory usage and inference latency relative to other quantization methods. These results demonstrate that carefully designed quantization acts as an effective regularizer, enabling the practical deployment of sophisticated LLM-based reasoning on edge devices. Code is available at: this https URL .

43. Reliable Reasoning with Large Language Models via Preference-Based Maximum Satisfiability

Authors: Pedro Orvalho , Marta Kwiatkowska , Guillem Alenyà , Felip Manyà
URL: https://arxiv.org/abs/2605.29687
Abstract:

Large Language Models (LLMs) excel at understanding natural language but struggle with optimisation tasks involving multiple constraints and user-defined preferences, which commonly arise in domains such as robotics. We propose a hybrid reasoning approach in which LLMs externalise reasoning through code generation. Given a natural language problem description, an LLM generates Python code that encodes user-defined constraints and preferences as a preference-based Maximum Satisfiability (MaxSAT) problem, which is then solved by an exact MaxSAT solver. To ensure correctness, solutions returned by the model-generated code are independently verified for feasibility and optimality against a canonical MaxSAT encoding, allowing for different encodings and multiple optimal solutions. We evaluate our approach using both open-source and closed-access LLMs on three families of preference-based reasoning tasks, and compare it against direct-answer, chain-of-thought, and program-of-thought baselines using the same models. While these baselines rarely produce feasible solutions, the MaxSAT-based pipeline achieves substantially higher acceptance rates, in some cases exceeding 80%. Our results demonstrate that LLM-driven code generation combined with preference-based MaxSAT enables solver-verifiable optimisation with respect to generated encodings, and substantially improves correctness under independently verified reference semantics.

Authors: Yunjin Qi , Zhaojun Jiang , Xuan Wu , Hanxi Pan , Yixuan Wang , Yanfang Liu , Xiang Ji , Churu Yu , Chunyuan Zheng , Yingze Chen , Jie He , Liuqing Chen , Zaifeng Gao
URL: https://arxiv.org/abs/2605.29685
Abstract:

As large language models (LLMs) are increasingly applied in social contexts such as emotional companionship and customer service, measuring their social intelligence has become critical to the quality and safety of human-AI interaction. However, existing social intelligence benchmarks lack a unified framework that organizes social abilities into a unified structure, and therefore cannot enable fine-grained diagnosis. To build the first holistic diagnostic evaluation grounded in social theory, we first construct a social intelligence framework through a literature review and multi-stage expert validation guided by psychometric principles. The resulting framework includes 4 categories and 11 dimensions, each further specified by fine-grained capability facets. Building on this framework, we introduce NICE (Norm, Interaction, Cognition, Experience), a diagnostic benchmark of 137 items operationalized through representative Chinese contexts. Across 5 frontier LLMs and a human reference group, models score higher in aggregate accuracy yet show a consistent weakness in Communication, which the framework localizes to 3 specific capability facets: multi-turn communication, nonverbal communication, and synchrony. NICE thus reframes social intelligence evaluation toward theory-grounded diagnosis of socially consequential weaknesses in LLMs.

45. Notation Matters: A Benchmark Study of Token-Optimized Formats in Agentic AI Systems

Authors: Lorenz Kutschka , Bernhard Geiger
URL: https://arxiv.org/abs/2605.29676
Abstract:

Large language models in Agentic AI systems consume tool schemas and execution results and emit tool invocations as structured data. The default language for that exchange, JSON, was designed for application-to-application interchange rather than token efficiency, so its structural elements impose substantial token overhead. Recent work proposes token-optimized alternatives such as TOON (Token-Oriented Object Notation) and TRON (Token Reduced Object Notation) as more compact replacements, but these formats have been evaluated only on isolated comprehension or generation tasks. Whether their token reductions hold inside end-to-end agentic loops therefore remains an open question. We evaluate TOON and TRON on four agentic benchmarks (BFCL, MCPToolBenchPP, MCP-Universe, StableToolBench) and five open-weight LLMs, decoupling input compression from output compression to measure comprehension and generation independently. TRON reduces tokens by up to 27% with accuracy within 14pp of the JSON baseline. TOON achieves up to 18% reduction at a similar 9pp accuracy cost, but additionally cascades on multi-turn parsing failures and collapses parallel tool-call output for most models.

46. GRASP: Gated Regression-Aware Skill Proposer for Self-Improving LLM Agents

Authors: Johannes Moll , Jean-Philippe Corbeil , Jiazhen Pan , Martin Hadamitzky , Daniel Rueckert , Lisa Adams , Keno Bressem
URL: https://arxiv.org/abs/2605.29668
Abstract:

LLM agents acting in structured environments fail in operational rather than conversational ways, and reliability depends on procedural knowledge of the environment. Prior self-improvement methods accumulate natural-language guidance without checking that each new item preserves previously correct behavior, so a note that fixes one trajectory can silently regress another. We introduce GRASP (Gated Regression-Aware Skill Proposer), which treats agent improvement as a sequence of edits to a bounded skill library, admitting each candidate only if it produces a net improvement on a balanced held-out probe under a hard regression budget. We evaluate GRASP across five base models (gpt-oss-120b, DeepSeek V4 Flash, Gemini 3.1 Flash Lite, GPT-4.1, GPT-5.4) on two FHIR-based clinical benchmarks. On MedAgentBench, GRASP lifts gpt-oss-120b from 40.6% to 88.8%, exceeds the strongest of five self-improvement baselines by 21.0 points, and improves every other base model by 17.2 to 40.3 points. Ablations attribute the gain to comparative proposal generation, the acceptance gate, and the hard regression budget rather than to skill writing itself, which without validation is no better than using no skills. The mechanism generalizes beyond the clinical domain, improving agents on three of four non-clinical environments and remaining flat only where the action space is open-ended. Frozen libraries transfer across models, where skills from a stronger model improve weaker executors beyond what they learn for themselves while the reverse does not, an asymmetry that no ungated baseline reproduces.

47. TRACE: Toulmin-based Reasoning Assessment through Constructive Elements for LLM CoT Evaluation

Authors: Yundong Kim , Heyoung Yang
URL: https://arxiv.org/abs/2605.29656
Abstract:

Evaluating open-ended outputs from large language models (LLMs) remains challenging due to the absence of ground truth. Existing metrics rely on final-answer accuracy or surface-level statistics, leaving the reasoning process itself unexamined. We introduce TRACE (Toulmin-based Reasoning Assessment through Constructive Elements), a metric that analyzes Chain-of-Thought (CoT) reasoning processes. Rather than judging outcomes, TRACE inspects how arguments are constructed by integrating Toulmin’s argumentation theory with Flavell’s metacognitive framework to assess reasoning structure. Experiments on 26.3K QA samples across 7 reasoning models show strong correlation with benchmark accuracy (r=0.74). Furthermore, TRACE is effective as a reinforcement learning reward signal, outperforming accuracy-only baselines. Together, these results indicate that logically sound reasoning leads to higher-quality answers. TRACE thus serves as a complementary metric for evaluating open-ended outputs. Code is available at this https URL .

48. PTCG-Bench: Can LLM Agents Master Pokémon Trading Card Game?

Authors: Dongdong Hua , Yifei Sun , Renhong Huang , Feng Gao , Chunping Wang , Yang Yang
URL: https://arxiv.org/abs/2605.29653
Abstract:

Given a strategically complex board game, human players can quickly learn to devise strategies after playing a few rounds. Autonomous agents require similar capabilities in realistic interactive environments, yet existing agent benchmarks often fail to fully capture such strategic and evolving decision-making scenarios. We present PTCG-Bench, a benchmark built on the Pok’{e}mon Trading Card Game (PTCG) that evaluates LLM agents at two complementary levels: (1) their decision-making performance within a single complex environment, and (2) their ability to self-evolving through accumulated experience. We further include a modular harness ablation to better interpret agent performance without conflating it with model capability. Our experiments show that, although LLM agents can achieve non-trivial gameplay performance, sustained and stable self-evolution remains challenging, and performance is sensitive to harness design. We hope that PTCG-Bench will facilitate future research on harness-aware and self-evolving agents in realistic interactive environments.

49. Think Fast, Talk Smart: Partitioning Deterministic and Neural Computation for Structured Health Text Generation

Authors: Kai-Chen Cheng , Haejun Han , David Q. Sun
URL: https://arxiv.org/abs/2605.29652
Abstract:

Large language models (LLMs) are increasingly being used to generate health text from structured records such as wearable time series, biomarkers, vitals, and care-management logs. For recurring health outputs, fluency is not enough: systems must remain faithful to source data, ground explanatory claims in available evidence, follow stated policies, emit machine-readable outputs, and run cheaply enough for repeated use. We ask which responsibilities in structured health generation should be deterministic computation rather than runtime LLM prompting. We introduce Think Fast, Talk Smart, a sleep-health insight pipeline in which deterministic code performs recurring analysis before one bounded LLM writer call. Across 280 user-nights and six models, achieves lower numeric error, lower instruction-compliance error, and lower end-to-end cost than structured zero-shot and few-shot one-call baselines. Layer replacement reveals contract-specific failures: LLM comparison raises numeric error, LLM ranking degrades policy selection, LLM attribution increases unsupported causal language, and an LLM-generated writer interface reintroduces errors even after upstream facts are deterministic. The results support a broader design rule: let code own recurring analysis, and let LLMs express verified facts within bounded interfaces.

50. LLM-Evolved Domain-Independent Heuristics for Symbolic AI Planning

Authors: Elliot Gestrin , Jendrik Seipp
URL: https://arxiv.org/abs/2605.29649
Abstract:

Heuristic search is the dominant paradigm in symbolic AI planning, and the strongest heuristics are the result of decades of work by planning researchers. Recent work has shown that large language models (LLMs) can design heuristics for individual planning domains, but no LLM-generated heuristic has so far worked on arbitrary planning tasks. In this paper, we use evolutionary search to produce the first LLM-generated domain-independent heuristics that exceed the hand-engineered state of the art. We let an LLM mutate parent heuristics written in C++, store candidates in a MAP-Elites archive keyed on informedness and speed and calculate fitness scores by blending coverage with solving time. To place the evolved programs in context, we additionally benchmark a broad set of hand-engineered heuristics on their informedness-speed tradeoff, which to our knowledge has not been done before. On unseen testing domains, our best evolved heuristic solves more tasks than even the strongest baseline, with our full heuristic suite spanning the Pareto frontier of said tradeoff. We also find that seeding evolution from the trivial blind heuristic outperforms seeding from the strong FF heuristic, even when the resulting program is itself an FF variant, and that LLM reasoning effort affects how often candidates compile much more than the quality of those that do. Because the evolved programs are plain C++, they slot into existing planners as drop-in replacements and inherit the soundness and completeness guarantees of the underlying search.

51. VikingMem: A Memory Base Management System for Stateful LLM-based Applications

Authors: Jiajie Fu , Junwen Chen , Mengzhao Wang , Aoxiang He , Maojia Sheng , Xiangyu Ke , Yifan Zhu , Yunjun Gao
URL: https://arxiv.org/abs/2605.29640
Abstract:

Large Language Models have revolutionized interactive applications; however, their finite context windows pose a critical data management challenge for maintaining stateful, long-term interactions. Existing memory approaches often rely on simplistic extraction methods that lead to incomplete memories or use rigid, single-purpose memory extraction prompts tailored to a single use case, such as chatbots. Consequently, they lack generalizability and perform poorly across diverse downstream tasks. To bridge this gap, we introduce the Memory Base, a novel data management paradigm for managing the persistent state of long-term interactions. It is characterized by three core principles: selective extraction of high-value memories from raw information streams; inherent statefulness and evolution, where memory content is progressively summarized, corrected, and temporally weighted to prioritize recent interactions; and a generalizable abstraction paradigm designed for robust transferability across diverse applications, including education, recommendation, and agent memory. Building on this foundation, we present VikingMem, an end-to-end Memory Base Management System implemented on the VikingDB vector engine. VikingMem materializes this paradigm through interconnected event and entity abstractions. It features event-centric memory extraction to selectively handle complex information streams, while entities are dynamically updated by events to achieve stateful evolution. Using temporal compression via a topic-wise timeline and time-weighted recall, the system progressively produces high-level summary memories, prioritizes recent items, and compresses and fades older ones. Extensive evaluations on long-term memory benchmarks demonstrate that VikingMem outperformes baselines by up to 30% in memory retrieval effectiveness while maintaining the low latency essential for interactive applications.

52. Beyond Attack Success Rate: Temporal Logit Observability for LLM Safety Failures

Authors: Junyoung Park , Sunghwan Park , Seongyong Ju , Jaewoo Lee
URL: https://arxiv.org/abs/2605.29629
Abstract:

Attack Success Rate (ASR) evaluates each jailbreak with a single yes/no label at the end of generation, telling us whether a failure happened but not how it unfolded. Two attacks that produce equally harmful outputs may have followed completely different paths, and ASR cannot tell them apart. We make those hidden paths observable from logits alone. Temporal Logit Observability (TLO) is a training-free diagnostic that watches a compliance-refusal margin during decoding and places each model-attack condition on a calibrated 2D plane. By design, this plane is most informative exactly where ASR is least informative: among attacks that succeed for genuinely different reasons. Across four aligned LLMs and three jailbreak paradigms, attacks with nearly identical ASR land at clearly different points on the plane: the same model can fail through different temporal patterns. The geometry matches refusal-direction probes from hidden states on most conditions, with one model showing the limit of our fixed-lexicon approach. A simple early-stop rule derived from TLO cuts successful jailbreaks by more than half, without false alarms on plain benign queries. Safety evaluation should report when and how a failure unfolds, not only whether it occurred. TLO makes the first two observable from logits alone.

53. Improving Collaborative Storytelling with a Multi-Agent Framework Based on Large Language Models

Authors: Arturo Valdivia , Paolo Burelli
URL: https://arxiv.org/abs/2605.29625
Abstract:

The topic of Co-creation, i.e., AI agents interacting with humans to generate outputs (e.g., art), has gained significant attention recently. However, most studies focus on adult-human interactions in a digital setting. This paper explores a novel ludic co-creation scenario involving children and Large Language Models (LLMs) interacting through a physical board game to create written stories. Our goal is to develop a multi-agent framework capable of producing high-quality narratives suitable for young players. At the core of our approach is an iterative Writer-Editor process in which one LLM generates stories while another evaluates them and provides feedback for refinement. Through a simulation study involving multiple LLMs, we show that this iterative interaction consistently improves the perceived quality of generated stories across successive loops. The results indicate that a small number of refinement steps may be sufficient to achieve high-quality outputs in interactive storytelling systems.

54. FinVerBench: Benchmark Validity and Calibration in Large Language Model Financial Statement Verification

Authors: Silu Panda
URL: https://arxiv.org/abs/2605.29586
Abstract:

We introduce FinVerBench, a benchmark and validity study for financial statement verification: determining whether a set of corporate financial statements is numerically consistent from the information shown to the model. FinVerBench is built from SEC 10-K XBRL filings for 43 S&P 500 companies and defines a four-category error taxonomy covering arithmetic, cross-statement linkage, year-over-year, and magnitude perturbations. We attempt fifteen contemporary LLM evaluations and report fourteen complete runs; a Gemini 2.5 Pro run is excluded from the main comparison because 40/108 gateway calls failed. All binary metrics exclude underdetermined positive instances whose perturbed line item is not rendered, leaving a 105-instance observable diagnostic subset (43 clean, 62 error-injected). Under the original guided-checklist prompt on the unrounded diagnostic subset, nine of fourteen complete LLM runs produce 95-100% false positives on clean statements, while one run achieves 0% observed false positives. Benchmark rendering choices materially affect measured recall: on a realistic rounded variant of the same observable subset, the calibrated model’s recall is 79.0% with 0% observed FPR, compared with 100.0% recall on the unrounded diagnostic variant. These results support a construct-validity conclusion rather than a final leaderboard: financial statement verification is not merely arithmetic detection, but calibrated judgment under incomplete observability, prompt-induced assumptions, and realistic numerical rendering. FinVerBench and all code are publicly available.

55. GPS-Enhanced Tourist Mobility Modeling with Seasonal Spatial Priors and LLM-Based Activity Chain Generation

Authors: Yifan Liu , Yanling Sang , Xishun Liao , Morgan Sun , Bo Yang , Zhiyuan Zhang , Chris Stanford , Haoxuan Ma , Jiaqi Ma
URL: https://arxiv.org/abs/2605.29578
Abstract:

Tourist mobility poses a distinct challenge for urban transportation planning. Unlike resident commuting, tourist travel is largely non-routine, attraction driven, and highly sensitive to trip purpose, travel season, and trip member composition. Existing approaches either measure aggregate tourist spatial patterns without generating individual schedules, or synthesize mobility without tourist specific structure such as trip duration conditioning, month varying attraction demand, and household co-travel rules. To address these challenges, we propose a four stage simulation framework combining month conditioned spatial priors derived from GPS and survey data, trip extent prediction from tourist demographics, distance feasible ward sequence assignment, and LLM-based activity chain generation under household and spatial constraints. GPS data are used only in privacy preserving aggregated form as month conditioned spatial priors, with no individual traces retained or exposed. Experiments on tourism in Tokyo demonstrate that the GPS based tourist cohort extraction recovers spatial visitation signatures consistent with survey references, and our framework produces demographically aligned synthetic schedules whose ward-level visitation shares align closely with both survey distributions and staypoint derived monthly visitation patterns. The results demonstrate the framework’s effectiveness as a geographically grounded, demographically aware approach to tourist mobility modeling.

56. DeepTool: Scaling Interleaved Deliberation in Tool-Integrated Reasoning via Process-Supervised Reinforcement Learning

Authors: Yang He , Xiao Ding , Bibo Cai , Yufei Zhang , Kai Xiong , Zhouhao Sun , Bing Qin , Ting Liu
URL: https://arxiv.org/abs/2605.29568
Abstract:

Tool-Integrated Reasoning (TIR) extends LLM capabilities by leveraging external environments. However, existing methods lack the deliberation during sequential tool invocation required for strategic planning and self-correction. While RL mitigates this, conventional approaches for Tool-Integrated Reasoning are hindered by sparse outcome-based rewards, failing to supervise intermediate reasoning steps and tool invocations. To address this, we propose DeepTool, a novel framework that scales deliberate thinking within the interleaved process of thinking, action, and observation at each turn. In DeepTool, we first introduce a synthesis pipeline that evolves extended thinking into interleaved trajectories, integrating adversarial perturbations to ensure robustness and self-correction. Secondly, we devise Process-Supervised Reinforcement Learning based on GRPO, which utilizes an Action-Centric Process Reward to reinforce intermediate interleaved thinking and enforce precise tool invocation at every turn. Extensive experiments demonstrate that DeepTool achieves superior performance, boosting Qwen2.5-7B significantly across six benchmarks (e.g., AIME24: 3.2% -> 40.4% and HMMT25: 0.0% -> 28.6%). Furthermore, the token cost-effectiveness analysis confirms the utility of interleaved thinking, demonstrating DeepTool’s optimal balance between performance and token efficiency.

57. ParaTool: Shifting Tool Representations from Context to Parameters

Authors: Zekai Yu , Qi Meng , Qizhi Chu , Yu Hao , Chuan Shi , Cheng Yang
URL: https://arxiv.org/abs/2605.29561
Abstract:

Tool calling extends large language models (LLMs) by enabling grounded interaction with external executable interfaces, thereby supporting environment-coupled problem solving. However, mainstream in-context learning (ICL) approaches typically incorporate detailed tool documentation and usage examples directly into the context. This results in substantial inference overhead and heightened risks of hallucination as the context length grows. Conversely, while tuning-based methods improve general tool-calling capabilities, they often fail to effectively internalize the specific details of previously seen tools, thereby retaining a dependency on in-context documentation. To address these limitations, we propose ParaTool, a framework that projects each tool into a dedicated, loadable set of parameters. By equipping a dynamic integration of these parameterized tools, the LLM can perform tool calling without relying on in-context documents or examples. Specifically, our approach consists of three stages: (1) parametric tool pre-training encapsulates the knowledge of different tools into independent parameter modules; (2) soft tool selection employs a gating network to dynamically weigh and aggregate relevant tool parameters; and (3) parametric tool fine-tuning jointly updates tool parameters to align the training and inference processes. Experiments on Stable ToolBench and BFCL demonstrate that ParaTool significantly outperforms strong ICL-based baselines, achieving superior performance while reducing computational complexity.

58. Battery-Sim-Agent: Leveraging LLM-Agent for Inverse Battery Parameter Estimation

Authors: Jiawei Chen , Xiaofan Gui , Shikai Fang , Shengyu Tao , Shun Zheng , Weiqing Liu , Jiang Bian
URL: https://arxiv.org/abs/2605.29560
Abstract:

Parameterizing high-fidelity “digital twins” of batteries is a critical yet challenging inverse problem that hinders the pace of battery innovation. Prevailing methods formulate this as a black-box optimization (BBO) task, employing algorithms that are sample-inefficient and blind to the underlying physics. In this work, we introduce a new paradigm that reframes the inverse problem as a reasoning task, and present Battery-Sim-Agent, the first framework to deploy a Large Language Model (LLM) agent in a closed loop with a high-fidelity battery simulator. The agent mimics a human scientist’s workflow: it interprets rich, multi-modal feedback from the simulator, forms physically-grounded hypotheses to explain discrepancies, and proposes structured parameter updates. On a systematically constructed benchmark suite spanning diverse battery chemistries, operating conditions, and difficulty levels, our agent significantly outperforms strong BBO baselines like Bayesian optimization in identifying accurate parameters. We further demonstrate the framework’s capability in complex long-horizon degradation fitting tasks and validate its practical applicability on real-world battery datasets. Our results highlight the promise of LLM-agents as reasoning-based optimizers for scientific discovery and battery parameter estimation.

59. Opt-Verifier: Unleashing the Power of LLMs for Optimization Modeling via Dual-Side Verification

Authors: Haoyang Liu , Jie Wang , Boxuan Niu , Xiongwei Han , Yian Xu , Mingxuan Ye , Zijie Geng , Fangzhou Zhu , Tao Zhong , Mingxuan Yuan , Jianye Hao
URL: https://arxiv.org/abs/2605.29556
Abstract:

Building mathematical optimization models is critical in operations research (OR), while it requires substantial human expertise. Recent advancements have utilized large language models (LLMs) to automate this modeling process. However, existing works often struggle to verify the correctness of the generated optimization models, without checking the rationality of the constraints and variables or the validity of solutions to the generated models. This hampers the subsequent verification and correction steps, and thus it severely hurts the modeling accuracy. To address this challenge, we propose a novel LLM-based framework with Dual-side Verification (Opt-Verifier) from both structure and solution perspectives, thereby improving the modeling accuracy. The structure-side verification ensures that the modeling structure of the generated optimization models aligns with the original problem description, accurately capturing the problem’s constraints and requirements. Meanwhile, the solution-side verification interprets and evaluates the solutions’ validity, confirming that the optimization models are logically and mathematically sound. Experiments on popular benchmarks demonstrate that our approach achieves over 20\% improvement in accuracy.

60. UI-KOBE: Knowledge-Oriented Behavior Exploration for Lightweight Graph-Guided GUI Agents

Authors: Yuxiang Chai , Han Xiao , Xinyu Fu , Jinpeng Chen , Rui Liu , Hongsheng Li
URL: https://arxiv.org/abs/2605.29534
Abstract:

Recent advances in mobile GUI agents have shown strong potential for automating mobile tasks, but most effective systems still depend on large vision-language models for screenshot understanding and long-horizon planning. Small GUI agents that can be deployed directly on mobile devices are more attractive for practical use, offering lower inference cost and better protection of sensitive on-device information. However, due to limited model capacity, such lightweight agents remain unreliable when planning and executing GUI tasks end-to-end from screenshots alone. We propose Knowledge-Oriented Behavior Exploration (\textbf{UI-KOBE}), a framework that improves lightweight mobile GUI agents with reusable app-specific graph knowledge. UI-KOBE first autonomously explores a mobile application and constructs an app knowledge graph, where nodes represent distinct UI states and edges represent executable transitions. At runtime, a lightweight GUI agent uses the graph as external guidance: given a user task and the current screenshot, it identifies the current graph node and selects among self-loop actions, neighboring transitions, task completion, or fallback free actions associated with that node. By supporting runtime decisions with app-specific graph guidance, UI-KOBE reduces the burden of end-to-end GUI planning and helps lightweight models perform mobile GUI tasks more effectively, offering a practical step toward efficient, interpretable, and privacy-conscious on-device GUI agents.

Authors: Kevin Wang , Anna Thöni , Benjamin Kempinski , Bobby Cheng , Jianzhu Yao , Benjamin Finch , Leon Guertler , Viraj Nadkarni , Yihan Jiang , Aliaksei Korshuk , Alexander Buyantuev , Ilya Makarov , Siyuan Wu , Yu-Chi Cheng , Yan-Ru Ju , Ti-Rong Wu , I-Hsuan Chu , Yu-Yu Yang , I-Chen Wu , Yitian Huang , Qinlu Cao , Yiheng Sun , Yuhong Dai , Hongkun Yao , Jingxuan Fu , Jiwei Zhang , Hao Liao , Mossimo Ebeling , Govind Arun , Sadhvik Bathini , Mihir S Arya , Avinash Anish , Aditya Ranjan , Kirtana Sunil Phatnani , Paval KS , Vrushali Mehta , Aravind S , Nikhil Arora , Tanya Upadhyay , Amol Bandagale , Yuan Lu , ChunEn Hsiao , YuTing Lin , Arvin Chung , Jerry John Thomas , Mathieu Laurière , Leshem Choshen , Yoram Bachrach , Pramod Viswanath , Maria Polukarov , Cheston Tan , Tal Kachman , Atlas Wang
URL: https://arxiv.org/abs/2605.29512
Abstract:

Large language models (LLMs) are increasingly deployed as interactive agents, yet their capacity for social and strategic reasoning over extended interaction remains poorly understood. Existing evaluations rely on static vignettes or single-game benchmarks that cannot capture the sustained, multi-faceted reasoning that real-world multi-agent settings demand. We introduce Mindgames, a multi-game arena and evaluation platform for LLM agents that operationalizes complementary reasoning demands relevant to ``theory of mind’’: belief attribution under hidden information, opponent modeling through repeated strategic interaction, cooperative inference under knowledge asymmetries, and sustained deception in social deduction. Built on TextArena, Mindgames provides a unified interaction interface, TrueSkill-based rating, and full trajectory logging across four game environments. We instantiate Mindgames through a 2025 competition cycle hosted at a major AI conference, which assessed 944 submitted agents from 76 teams across four games: Colonel Blotto, Iterated Prisoner’s Dilemma, Codenames, and Secret Mafia. Our analysis surfaces both agent-level and evaluation-level limitations: brittle rule adherence remains a major bottleneck, top-performing systems repeatedly rely on explicit structural scaffolding, and leaderboard validity differs sharply across environments. In particular, failure-heavy environments can reward robustness to opponent errors as much as strategic ability, with Secret Mafia exhibiting a pronounced error-survival confound in this cycle. We release a dataset of 29,571 multi-agent games with turn-level observations, actions, and rewards, together with MG-Ref, a deterministic offline tournament protocol that scores new agents against a frozen reference pool of top-ranked, low-error Stage~II submissions under the same error-attribution lens used in this analysis.

62. The Curse of Helpfulness: Inverse Scaling Law in Robustness to Distractor Instructions via DistractionIF

Authors: Zeli Su , Zhankai Xu , Tianlei Chen , Longfei Zheng , Xiaolu Zhang , Jun Zhou , Wentao Zhang
URL: https://arxiv.org/abs/2605.29491
Abstract:

Large Language Models (LLMs) are increasingly deployed in agentic and retrieval-augmented generation (RAG) systems, where they must execute user-specified tasks over externally provided reference text. In practice, such context is often unstructured and contaminated with benign but instruction-like semantic noise, such as editorial comments and system traces, which should be treated strictly as data. We introduce DistractionIF, a benchmark designed to evaluate robustness against such distractor instructions in reference text. Across a broad range of models, we observe a consistent inverse scaling phenomenon: larger models are often less robust, with performance dropping by up to 30 points as scale increases. Mechanistically, our perplexity analysis reveals that scaling erodes the probabilistic boundary between robust and distracted behaviors, making models increasingly prone to over-interpreting noise as instructions. To address this, we demonstrate that reinforcement learning, specifically Group Relative Policy Optimization (GRPO), can restore this boundary, improving robustness by up to 15.5% without compromising general instruction-following capability. Our findings highlight a critical instruction-following robustness gap in reference-grounded tasks and establish reinforcement learning as a promising path for enforcing strict data-instruction separation at scale.

63. CrystalXRD-Bench: Benchmarking Vision-Language Models for XRD Peak Indexing Across Diverse Crystalline Materials

Authors: Chengliang Xu , Xiaogang Li , Peiyao Xiao , Beng Wang , Hu Wei , Bing Zhao
URL: https://arxiv.org/abs/2605.29446
Abstract:

Miller-index identification from powder XRD patterns requires capabilities untested by existing multimodal benchmarks: the model must read a narrow peak location from a rendered scientific curve and then connect that observation to multi-step crystallographic reasoning. We introduce CrystalXRD-Bench, a 250-sample benchmark built from 10 public crystallographic databases for a single task: recover the full set of HKLs contributing to the highest-intensity peak in an XRD pattern. Each sample pairs the rendered XRD image with the source CIF text and chemical formula, so visual extraction errors and reasoning errors can be examined side by side. We evaluate seven vision-language models. The best Jaccard score is 0.5888 (GPT-5.4) with an exact-match rate of 37.6%, yet six of seven models remain below Jaccard 0.50; the task is far from solved. Error patterns vary systematically: double-peak cases are especially brittle, recall-heavy models gain coverage by over-predicting HKLs, and access to CIF text does not close the gap in crystallographic calculation. Alongside model rankings, the benchmark identifies the conditions under which current VLMs fail on quantitative scientific figures. All data and evaluation code will be publicly available.

64. Towards Human-Like Interactive Speech Recognition With Agentic Correction and Semantic Evaluation

Authors: Zixuan Jiang , Yanqiao Zhu , Peng Wang , Qinyuan Chen , Xinjian Zhao , Xipeng Qiu , Wupeng Wang , Zhifu Gao , Xiangang Li , Kai Yu , Xie Chen
URL: https://arxiv.org/abs/2605.29430
Abstract:

Automatic speech recognition (ASR) is a core component of human–computer interaction and an increasingly important front-end for LLM-based assistants and agents. However, most current ASR systems still follow a single-pass paradigm, which is poorly aligned with human communication, where misunderstandings are resolved through iterative clarification and refinement. This mismatch makes it difficult to correct meaning-critical errors once they occur. Meanwhile, token-level metrics such as WER or CER cannot adequately reflect such a problem. To address these limitations, we formulate \emph{Interactive ASR} as a multi-turn refinement task and propose \textbf{Agentic ASR}, a closed-loop framework that combines a single-pass ASR front-end with semantic correction, intent routing, and reasoning-based editing. We further introduce the \textbf{Sentence-level Semantic Error Rate} ($S^2ER$), an LLM-based semantic evaluation metric, together with an \textbf{Interactive Simulation System} for scalable and reproducible benchmarking. Experiments on multilingual, named-entity-intensive, and code-switching benchmarks show that iterative interaction consistently reduces semantic errors, with much larger gains in $S^2ER$ than in conventional token-level metrics. Human–AI alignment and ablation studies further validate the reliability of the semantic judge and the robustness of the proposed framework. The code is available at: this https URL and the live demo is available at this https URL

65. When Does Persona Prompting Actually Help? A Retrieval and Metric Analysis of Expert Role Injection in LLMs

Authors: Shuai Xiao , Su Liu , Weikai Zhou , Jialun Wu , Xinjie He , Zhiyuan Lin , Qiyang Xie
URL: https://arxiv.org/abs/2605.29420
Abstract:

Persona prompting is widely used to steer large language models, yet its practical value remains unclear. Prior work often evaluates persona prompting using aggregate scores, making it difficult to determine whether expert-role prompting consistently improves response quality or instead changes responses along different quality dimensions. We study this question through a controlled comparison of four prompting conditions across 1,140 open-ended questions spanning 38 expert roles and six domains: no role prompt, a generic domain-expert prompt, embedding-based role retrieval, and a hybrid retrieval method combining embedding search with LLM-based role selection. Aggregate results show only small overall differences between conditions. However, metric-level analysis reveals a consistent tradeoff that aggregate averages obscure: role prompting systematically increases expertise depth while reducing clarity. These effects are highly conditional rather than universal. Role prompting performs best on advisory questions and in domains such as medicine and psychology, where structured expert framing and risk communication are intrinsically valuable. In contrast, baseline prompting performs better on conceptual and explanatory questions in finance, legal, science, and technology domains, where concise plain-language explanation is more important. We further show that hybrid retrieval significantly improves over embedding-only role selection, although better role retrieval does not eliminate the broader expertise-depth versus clarity tradeoff. Overall, our findings suggest that persona prompting primarily reshapes response characteristics rather than broadly improving capability, and that multi-metric evaluation is necessary for understanding its effects.

66. Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization

Authors: Zhihao Liu , Yifan Wu , Jian Lou , Di Wang , Yuxi Zhou , Yuke Hu
URL: https://arxiv.org/abs/2605.29396
Abstract:

Safety alignment for large language models (LLMs) aims to reduce harmful or unsafe behavior while preserving general utility. However, recent findings reveal that alignment effects can be fragile: lightweight post-alignment manipulations, such as parameter noise, activation noise, or quantization, can easily weaken the intended safety behavior. Prior efforts to improve robustness have primarily focused on data curation, modified alignment objectives, and safety-critical parameter identification, leaving the role of the optimizer itself largely unexplored. In this paper, we are the first to study the robustness of safety alignment from the perspective of the base optimizer. This optimizer-centric view naturally points to zeroth-order optimization, which provides a robustness-oriented signal by evaluating safety alignment under perturbations. Based on this insight, we propose a hybrid framework that first performs standard first-order safety alignment and then applies zeroth-order refinement to improve robustness. Both theoretically and empirically, we show that only a few zeroth-order refinement steps can enhance robustness while preserving safety alignment. We further improve the efficiency of zeroth-order refinement by exploiting its inherent perturbation-based evaluations to estimate layer-wise robustness sensitivity, enabling the refinement process to concentrate updates on robustness-critical layers with modest training overhead.

67. EvoMD-LLM: Learning the Language of Species Evolution in Reactive Molecular Dynamics

Authors: Zhichen Tang , Zhengzheng Dang , Yulin Chen , Jixin Wu , Haiwen Li , Yanming Wang
URL: https://arxiv.org/abs/2605.29394
Abstract:

While large language models (LLMs) excel at static scientific reasoning, they struggle to model the temporal structure of dynamic physical processes. We present EvoMD-LLM (Evolutionary Molecular Dynamics Large Language Model), a framework that reformulates species-level molecular dynamics as a symbolic temporal language modeling problem. Reactive MD trajectories are discretized into sequences of molecular events, where each token represents a chemical species augmented with its persistence duration, enabling standard autoregressive LLMs to learn compositional evolution over time through efficient fine-tuning. A key component of EvoMD-LLM is temporal scaffolding, which treats event duration as an explicit linguistic token and serves as a structured inductive bias, significantly reducing invalid or hallucinated molecular outputs compared to conventional sequence modeling approaches. We evaluate EvoMD-LLM on multiple temporal prediction tasks, achieving up to 66.14% accuracy and consistently outperforming sequential neural networks and language-based baselines. Beyond quantitative improvements, we qualitatively observe that the model is capable of generating interpretations for its own predictions by incorporating relevant chemical knowledge, even though it was not explicitly supervised with paired trajectory-explanation data. These results demonstrate that symbolic temporal language modeling provides an effective framework for grounding LLMs in dynamic physical simulations.

68. Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet

Authors: Adly Templeton , Tom Conerly , Jonathan Marcus , Jack Lindsey , Trenton Bricken , Brian Chen , Adam Pearce , Craig Citro , Emmanuel Ameisen , Andy Jones , Hoagy Cunningham , Nicholas L Turner , Callum McDougall , Monte MacDiarmid , Alex Tamkin , Esin Durmus , Tristan Hume , Francesco Mosconi , C. Daniel Freeman , Theodore R. Sumers , Edward Rees , Joshua Batson , Adam Jermyn , Shan Carter , Chris Olah , Tom Henighan
URL: https://arxiv.org/abs/2605.29358
Abstract:

We demonstrate that sparse autoencoders can extract interpretable features from Claude 3 Sonnet, a production-scale language model, addressing the open question of whether dictionary learning methods scale beyond small transformers. We trained sparse autoencoders with up to 34 million features on the model’s middle layer residual stream, using scaling laws to guide hyperparameter selection. The resulting features are multilingual and multimodal (generalizing to images despite text-only training), respond to both concrete instances and abstract discussions of concepts, and can be used to steer model behavior in ways consistent with their interpretations. We find features corresponding to famous entities and locations, as well as more abstract concepts like sarcasm or errors in code. We also identify features relevant to ways in which language models might cause harm–including features representing deception, power-seeking, sycophancy, and bias–and show that these causally influence model outputs when manipulated. Additionally, we conduct analyses of feature interpretability, geometry, and computational function. However, significant limitations remain: our suite of features is incomplete, and we lack rigorous methods for evaluating whether our features faithfully capture model computations.

69. PassNet: Scaling Large Language Models for Graph Compiler Pass Generation

Authors: Yiqun Liu , Yingsheng Wu , Ruqi Yang , Enrong Zheng , Honglei Qiu , Sijun He , Tai Liang , Jingjing Wu , Yuhan Zhou , Yiwei Zhang , Dongyan Chen , Weihan Yi , Xinqi Li , Siqi Bao
URL: https://arxiv.org/abs/2605.29357
Abstract:

Modern tensor compilers such as TorchInductor deliver substantial speedups on mainstream models, yet face a systematic performance ceiling on long-tail workloads – our profiling shows that 43% of real-world subgraphs experience end-to-end slowdowns under default compilation. While LLMs offer a path toward automated optimization, existing efforts focus on standalone kernel generation. We argue that pass generation – where LLMs author structured graph transformations that integrate directly into compiler pipelines – is the more appropriate abstraction. We propose PassNet, the first large-scale ecosystem for LLM-based compiler pass generation, comprising: (1) PassNet-Dataset, over 18K unique computational graphs from 100K real-world models; and (2) PassBench, 200 curated long-tail fusible tasks (comprising 2,060 subgraphs in total) evaluated under the Error-aware Speedup Score (ES_t) – a metric unifying correctness, stability, and performance – with layered integrity defenses against systematic LLM exploitation. Experiments reveal that PassBench is both highly discriminative and genuinely unsaturated: the best frontier model trails TorchInductor by 37% in aggregate, yet on individual subgraphs LLMs achieve up to 3x speedup over the same compiler – indicating that the bottleneck is consistency, not capability. Fine-tuning a small model on merely ~4K PassNet trajectories yields a 2.67x improvement approaching frontier-model performance, demonstrating substantial headroom and validating PassNet as live training infrastructure for advancing LLM-driven compiler optimization. All data, benchmarks, and tooling are publicly available.

70. ConMoE: Expert-Pool Consolidation via Prototype Reassignment for MoE Compression

Authors: Yilun Yao , Jiaming Pan , Elsie Dai , Peizhuang Cong , Yaoming Li , Tong Yang
URL: https://arxiv.org/abs/2605.29350
Abstract:

Mixture-of-Experts (MoE) language models reduce per-token computation but still require storing and serving all experts, making deployment memory-intensive. Existing post-training compression methods mainly shrink this cost by pruning experts or merging their weights. We formulate post-training MoE compression as expert-pool consolidation: retaining a smaller set of pretrained experts as reusable prototypes and deterministically remapping each original expert reference to one selected prototype. This view separates the reduced expert pool from the reuse structure that represents the original expert slots, and allows prototype sharing within local layer scopes while preserving the original router interface. We propose ConMoE, a train-free prototype remapping framework that selects retained experts using calibration-based contribution and replaceability signals, then redirects original expert calls to the selected prototypes without weight updates or post-compression fine-tuning. Experiments on three pretrained MoE language models show that ConMoE matches or outperforms strong pruning and merging baselines in several settings, achieving the best average score on deepseek-moe-16b-base at both 25% and 50% routed-expert reduction, while remaining competitive on Qwen3-30B-A3B and OLMoE-1B-7B-0125. Ablations indicate that deterministic reassignment is the most stable component, whereas broader cross-layer sharing and post-hoc weight fusion are model-dependent.

71. Entropy-KL Divergence-based Token Masking: A Novel Approach for Selective Fine-tuning of Large Language Models

Authors: Qi Liu , Mingdi Sun , Yongyi He , Zhi Zheng , Tong Xu , Yi Zheng , Zhefeng Wang , Enhong Chen
URL: https://arxiv.org/abs/2605.29303
Abstract:

Supervised fine-tuning (SFT) followed by reinforcement learning (RL) has become a standard post-training paradigm for large language models. This paradigm provides a cold-start for RL exploration, avoiding the inefficiency of pure RL where on-policy sampling yields insufficient positive samples. However, in practice, existing approaches often use a small amount of data for SFT initialization compared to the RL phase, which can cause the model to fit the limited samples and shift away from its pre-trained distribution. This distribution shift impedes the model’s ability to effectively explore during subsequent RL training. To address this challenge, we propose that in low-data regimes, SFT should prioritize activating task-relevant capabilities rather than memorizing specific content. Along this line, we propose EKSFT (Entropy-KL Selective Fine-Tuning), which selectively masks tokens that exhibit either high entropy or high KL divergence from a reference model. By excluding these high-uncertainty, distribution-shifting tokens from imitation, EKSFT injects task-specific knowledge while preserving the integrity of the model’s pre-trained distribution. Empirical evaluations on mathematical reasoning benchmarks demonstrate that EKSFT consistently outperforms standard SFT. Further RL fine-tuning from the EKSFT model yields consistently better post-RL performance, indicating improved exploration for the RL stage. Our codes and datasets are available at this https URL .

72. Diagnosing Harmful Continuation in Answer-Correct Long-CoT Training Traces

Authors: Chen He , Yuhao Wu , Lei Wang , Wenxuan Zhang , Fumin Shen
URL: https://arxiv.org/abs/2605.29288
Abstract:

Long chain-of-thought (CoT) traces are widely used as supervision for reasoning-oriented LLM SFT, yet answer-correct traces can still lead to markedly different fine-tuning outcomes. We study post-conclusion continuation in answer-correct long-CoT data: a continuation where the answer appears sufficiently supported, but the trace continues with additional reasoning that remains in the supervised target. To test its training effect, we use a delete-only editor to construct answer-preserving suffix removal and compare CoT-based SFT on the original and processed traces. We observe improved SFT outcomes after removing the editor-identified post-conclusion continuation, suggesting that this continuation is harmful to training in our setting. We therefore refer to this empirically supported phenomenon as harmful continuation. Beyond this intervention, we further characterize the removed post-conclusion continuation through uncertainty and hidden-state progress. We observe persistent local uncertainty together with weakened terminal-directional progress, forming an uncertainty–geometry mismatch. Finally, we instantiate Harmful Continuation Cut (HCC), a lightweight boundary proxy that approximates the editor-identified post-conclusion continuation boundary.

73. CoHyDE: Iterative Co-Training of LLM Rewriter & Dense Encoder for Tool Retrieval

Authors: Vaishali Senthil , Ashutosh Hathidara , Sebastian Schreiber
URL: https://arxiv.org/abs/2605.29271
Abstract:

Tool retrieval over large API catalogs is a core bottleneck for LLM agents: user queries arrive in colloquial, often underspecified language, while the catalog uses technical API vocabulary that no fixed encoder can bridge on its own. The two dominant training approaches, contrastive encoder fine-tuning and HyDE-style query expansion with a frozen LLM, address this problem from opposite ends and fail in complementary directions: the fine-tuned encoder excels when the query’s surface form already matches the catalog but collapses when it does not, while zero-shot HyDE is more robust to underspecified queries yet generates catalog-unaware hypothetical descriptions that degrade retrieval when queries are well-formed. We introduce CoHyDE, an iterative procedure that trains the dense encoder and the LLM rewriter as a single co-evolving system: the encoder is retrained with InfoNCE on catalog-style hypothetical descriptions produced by the rewriter, and the rewriter is preference-aligned via DPO against the encoder’s retrieval scores, with both sides warm-started on the tool catalog before the loop begins. On a ~10k tool subset of the ToolBench catalog, three rounds of CoHyDE improve over the strongest single-component baseline by +2.5 pp NDCG@5 on standard queries and +6.3 pp on held-out vague queries, with gains as large as +8 pp on the hardest vague tier. Ablations confirm that co-training is the key ingredient: using either component in isolation fails to match CoHyDE on both well-formed and vague queries, with losses of up to -8 pp on vague queries.

74. Indexing the Unreadable: LLM-Native Recursive Construction and Search of Service Taxonomies

Authors: Wei Zheng , Yang Yan , Yiyang Shao , Jinyang Li , Zeze Chang , Yukuang Jia , Qiming Mao , Chihyung Wang , Jingbin Zhou
URL: https://arxiv.org/abs/2605.29270
Abstract:

The era of the Internet of Agents (IoA) is taking shape: LLM agents are expected to fulfill user goals by orchestrating fast-growing populations of Model Context Protocol (MCP) servers, Agent-to-Agent (A2A) endpoints, reusable skills, and other LLM-callable services. Yet LLMs face a structural mismatch with this regime: effective context is a scarce resource that does not scale with the number of services. Concatenating thousands of service descriptions into a prompt overflows the context window, and even when the window is large enough, models systematically under-attend to information in the middle of long inputs, the well-documented Lost-in-the-Middle phenomenon. This is fundamentally a question of context management for service discovery. To address this, we propose an LLM-native progressive-disclosure scheme and its concrete instantiation, A2X (Agent-to-Anything service discovery): an LLM-driven pipeline that automatically organizes the registered services into a hierarchical taxonomy and walks it layer by layer at query time, so that every LLM call sees only a small candidate set highly relevant to the user query. This decouples effective-context scarcity from registry size and significantly reduces token consumption while improving retrieval accuracy. Compared to full-context dumping, A2X achieves a 6.2-point Hit Rate gain at one-ninth the prompt-token cost; compared to the state-of-the-art open-source embedding-based baseline, A2X improves Hit Rate by more than 20 points.

75. Harmonizing Real-Time Constraints and Long-Horizon Reasoning: An Asynchronous Agentic Framework for Dynamic Scheduling

Authors: Shijie Cao , Yuan Yuan , Jing Liu
URL: https://arxiv.org/abs/2605.29262
Abstract:

The Dynamic Flexible Job Shop Scheduling Problem (DFJSP) necessitates a trade-off between instant reaction to stochastic disturbances and global optimization of production goals. Conventional priority rules are insufficiently flexible to handle complex disruptions, whereas learning-based approaches often compromise interpretability or fail to generalize across problem scales. Although Large Language Models (LLMs) offer advanced reasoning capabilities to bridge this gap, their substantial inference latency is incompatible with the millisecond-level decision cycles of industrial control systems. To resolve this conflict, we introduce RACE-Sched, an asynchronous agent-based framework that decouples policy execution from logical reasoning via a dual-stream architecture. The Reactive Stream executes low-latency symbolic heuristics to enable real-time dispatching, while the parallel Deliberative Stream leverages an LLM to synthesize, validate, and evolve these rules. Candidate rules undergo rigorous testing in a sandbox and are deployed via atomic updates, ensuring safety without blocking the control loop. Additionally, a semantic rule repository indexes validated heuristics for retrieval-based initialization which enhances transferability across problem scales. Extensive evaluations on GEN-Bench, MK-Bench, and JMS-Bench demonstrate that RACE-Sched outperforms leading Deep Reinforcement Learning and other LLM-based baselines. This approach harmonizes real-time constraints with long-horizon reasoning to achieve superior solution quality and robust adaptation to dynamic events.

76. Provably Secure Agent Guardrail

Authors: Benlong Wu , Weiming Zhang , Kejiang Chen , Han Fang , Nenghai Yu
URL: https://arxiv.org/abs/2605.29251
Abstract:

As large language models transition from bounded generative engines to agents with expansive execution privileges, AI going out of control precipitates a fundamental crisis in artificial intelligence security. Existing defense architectures heavily rely on empirical semantic guardrails and probabilistic large model adjudicators, mechanisms that fail to provide deterministic security lower bounds when facing complex semantic symbol decoupling attacks. To overcome this empirical semantic guardrail dilemma, this paper proposes a new security paradigm for agents based on the fundamental limitations of logical reasoning. Based on this paradigm, we further introduce an executable Proof-Constrained Action (ePCA) framework with a neural symbolic isolation architecture. This framework abandons semantic trust in natural language, forcing agents to losslessly formalize their intentions into first-order logical mathematical constraints before performing physical operations. Empirical evaluations of macroscopic and microscopic two-dimensional dynamic adversarial systems demonstrate that our formal verification mechanism achieves zero attack success rate and zero false positive rate across the evaluated scenarios, with extremely low computational latency. This research provides a conditional formal foundation under explicit system assumptions and an engineering paradigm for constructing the underlying defense foundation for future intelligent systems.

77. DenseSteer: Steering Small Language Models towards Dense Math Reasoning

Authors: Yang Ouyang , Shuhang Lin , Jung-Eun Kim
URL: https://arxiv.org/abs/2605.29247
Abstract:

Large language models (LLMs) demonstrate strong chain-of-thought (CoT) reasoning abilities, while smaller models (<= 3B parameters) significantly underperform on multi-step reasoning tasks. Based on empirical analyses of the Qwen-2.5 model family on math reasoning benchmarks, we find that more proficient reasoning is associated with fewer reasoning steps but higher information density per step, a property we term Dense Reasoning. Motivated by this observation, we propose DenseSteer, a training-free inference-time steering framework that enhances small-model reasoning by modulating internal representations toward dense reasoning patterns. Experiments show that our method yields consistent accuracy improvements without increasing token-level Negative Log-Likelihood, highlighting dense reasoning as an effective structural approach to mathematical problem solving.

78. Rethinking Literature Search Evaluation: Deep Research Helps, and Human Citation Lists Are Not a Ground Truth

Authors: Gaurav Sahu , Laurent Charlin , Christopher Pal
URL: https://arxiv.org/abs/2605.29234
Abstract:

We study large-scale literature search from two complementary angles: improving the retrieval pipeline, and stress-testing the human reference list as an evaluation target. First, we implement a Deep Research pipeline that processes the full query paper and expands the retrieved results breadth-first along their bibliographies, and show that it substantially outperforms vanilla API-only search, raising recall on RollingEval-Jun25 (a 250-paper literature-search benchmark) from below 20% to above 80%. Second, we use a neutral LLM-as-a-judge to determine if human references are sound ground truth for the task. We find significant limitations: only 51% of human citations are judged moderately relevant or higher, against 86–88% for the strongest AI-based re-rankers. We study this gap on the OpenAlex co-authorship graph, finding that humans are 2.5x more likely than the best AI re-rankers to cite a direct collaborator. Together, our results argue against single-axis literature-search evaluation: recall, topical-relevance scoring, ranked-list diversity, and a co-authorship-distance diagnostic each measure complementary properties of citation quality and should be reported jointly.

79. Tailoring the Curriculum: Student-Centered Reasoning Distillation via Dynamic Data-Model Compatibility

Authors: Jiahao Huang , Fei Cheng , Junfeng Jiang , Akiko Aizawa
URL: https://arxiv.org/abs/2605.29229
Abstract:

Reasoning distillation transfers complex reasoning abilities from large language models (LLMs) to smaller ones, yet its success depends on how well the training data align with the student model. This paper introduces the Data-Model Compatibility (DMC) metric, which can be used to assess the suitability of a dataset for reasoning distillation on a student model. DMC provides an assessment by jointly considering data quality, relative difficulty, and student capability. We validated the effectiveness of DMC from two perspectives: (1) DMC exhibits a strong correlation with reasoning distillation performance; and (2) using DMC as the criterion for data selection leads to improved reasoning distillation performance. Both findings are consistently demonstrated across multiple student models and tasks. Moreover, since the DMC of each dataset dynamically changes during training, our experiments demonstrate that dynamically selecting datasets based on DMC can further enhance performance.

80. BenchTrace: A Benchmark for Testing Reflection Ability and Controlled Evolution in LLM Agents

Authors: Jiahao Huang , Fei Cheng , Junfeng Jiang , Zefan Yu , Akiko Aizawa
URL: https://arxiv.org/abs/2605.29225
Abstract:

Self-evolving agents improve over time by reflecting on past failures, but existing evaluation is limited in two ways: it measures only task scores, leaving reflection quality unknown, and it relies on agents’ own episode runs, offering no mechanism to target specific failure patterns. We present \textbf{BenchTrace}, a benchmark for evaluating self-evolution ability in LLM agents. BenchTrace is built on a snapshot-reflection dataset of 1,821 annotated episodes spanning six diverse tasks, and comprises a \textbf{Reflection Evaluation} that probes failure identification through targeted QA tasks, and an \textbf{Evolution Evaluation} that tests whether past failure experience translates into avoidance behavior in a controlled self-evolution simulation. Building on BenchTrace, we propose \textbf{failure avoidance rate (FAR)}, a new evaluation metric measuring the fraction of test cases in which the agent successfully avoids the target failure instance. Experiments with Qwen3-32B and GPT-4.1 reveal that both models fall below a 30\% end-to-end pass rate on reflection evaluation, with diagnosis as the primary bottleneck. Evolution evaluation shows that self-evolution methods generally improve FAR over the non-evolving baseline, but agents forget early lessons as noise episodes accumulate, and agents fail to generalize their reflections beyond the specific context, causing negative transfer across task contexts. Our correlation analysis further reveals that only a fully correct reflection is strongly associated with higher FAR. BenchTrace exposes concrete limits of current self-evolution approaches and provides a controlled, model-agnostic framework for targeted evaluation.

81. GTA: Generating Long-Horizon Tasks for Web Agents at Scale

Authors: Tenghao Huang , Kung-Hsiang Huang , Prafulla Kumar Choubey , Yilun Zhou , Muhao Chen , Jonathan May , Chien-Sheng Wu
URL: https://arxiv.org/abs/2605.29218
Abstract:

Web agents, which couple language models with browsing and tool-use capabilities, show promise as open web assistants. Yet progress is increasingly limited by the lack of scalable, process-level supervision. Existing benchmarks are largely manually constructed, providing only coarse start-goal annotations without intermediate trajectories, while recent automatic generation efforts remain expensive, biased, and shallow. These limitations prevent reliable training and evaluation of agents that must generalize to realistic, multi-hop, cross-page tasks. We introduce a scalable framework, GTA, that integrates crawling, retrieval-based seeding, in-context generation, and automated quality control to produce realistic tasks paired with executable trajectories. This design decouples crawling from generation for greater efficiency, grounds tasks in the site graph to enforce compositionality, and ensures dense supervision through deterministic replays and systematic validation. We instantiate the pipeline on over 50 websites covering e-commerce, government, forums, and news, with multilingual and multi-hop coverage. The resulting benchmark reveals a significant human-agent performance gap and enables detailed diagnostics. Our contributions are three-fold: (i) formalizing multi-hop web-agent task generation, (ii) proposing an efficient and validated pipeline for automatic data creation, and (iii) releasing a dynamic benchmark with reproducible evaluation.

82. ReasonOps: Operator Segmentation for LLM Reasoning Traces

Authors: Daniel Lee , Owen Queen , James Zou
URL: https://arxiv.org/abs/2605.29192
Abstract:

Chain-of-thought traces from large reasoning models can span tens of thousands of tokens, yet we lack a vocabulary for describing their internal structure. Previous methods developed to analyze chain-of-thought traces are either too rigid or not expressive enough, failing to capture features across domains and models. To remedy this, we develop ReasonOps, an unsupervised, expressive method for annotating chain-of-thought traces, providing succinct universal operators. Using ReasonOps, we analyze 44,662 traces from 12 thinking LLMs spanning 6 families across 8 reasoning benchmarks and discover that they share a common compositional structure: 7 recurring reasoning operators – discourse-level moves such as backtracking, inferring, and hypothesizing – that emerge from unsupervised clustering of sentence-initial 3-token pivots. These operators appear across every model family and benchmark domain, confirmed by three independent LLM judges who classify held-out samples at 70 -76% accuracy. We analyze the structure of operators on easy vs. hard problems, revealing that reflective operators are more helpful on hard problems and harm performance on easy problems. Operator sequences are highly model-identifying: a classifier trained on operator distributions alone recovers the source model with macro-AUC, revealing that each model family has a distinctive reasoning fingerprint. Structural operator features predict within-problem answer correctness well above baselines. Classifiers built on these operators reach WP-AUC and on AIME specifically. ReasonOps further enables early quality estimation well before the trace completes: we predict at WP-AUC for only 50% of the trace. The ReasonOps pipeline is unsupervised and annotation-free, enabling deep insights into LLM reasoning traces as well as strong downstream results on model identification and correctness prediction.

83. Better Later Than Sooner: Neuro-Symbolic Knowledge Graph Construction via Ontology-grounded Post-extraction Correction

Authors: Lorenzo Loconte , Timothy Hospedales , Cristina Cornelio
URL: https://arxiv.org/abs/2605.29168
Abstract:

Question answering (QA) is a core challenge in AI, particularly for complex queries requiring multi-hop reasoning across documents, or symbolic operations like aggregation or exhaustive listing. Retrieval-augmented generation has become the dominant approach to QA, with recent graph-based variants addressing part of these issues by organizing knowledge to better support compositional questions. However, most textual graph-based RAG methods still lack the structure needed for symbolic operations useful to answer complex questions reliably. This motivates symbolic graph-based approaches, which extract knowledge graphs (KGs) whose relations are logic predicates that enable SQL-like querying. Yet these pipelines typically use LLMs for KG extraction, which can introduce consistency issues, where extracted facts may violate commonsense ontology constraints. We propose a neuro-symbolic framework for ontology-grounded KG construction combining open-domain extraction, embedding-based canonicalization of types and predicates, and targeted LLM-based correction of ontology violations. By deferring corrections to a post-extraction stage, our method avoids repeated LLM calls, substantially reducing token usage while improving KG consistency and preserving downstream QA quality. Finally, we show that the extracted KGs are well suited for symbolic querying by measuring the occurrence of SPARQL graph patterns.

84. The Confidence Shortcut: A Reasoning Failure Mode of Masked Diffusion Models

Authors: Dueun Kim , Albert No
URL: https://arxiv.org/abs/2605.29123
Abstract:

Masked diffusion language models (MDMs) uniquely support any-order generation, with confidence-based decoding currently serving as the de facto standard inference policy. To optimize for this, recent training schemes attempt to align training mask patterns directly with those observed during generation. However, we argue that confidence-based decoding is inherently misaligned with the logical-flow trajectories required for complex reasoning, and that confidence-aligned training actively entrenches this misalignment. We make this concrete using multi-digit addition, where the decoding strategy prematurely predicts locally easy digits before resolving their long-range dependencies, producing high-confidence errors on challenging inputs. While traditional random masking keeps the failure rate low on this challenging tail, confidence-aligned training amplifies the error rate by an order of magnitude. Across five distinct reasoning tasks, this same pattern emerges with task-dependent severity: confidence-based decoding induces failures on highly complex inputs, and confidence-aligned training exacerbates them. In contrast, random masking – despite its perceived inefficiency – robustly preserves the reasoning-trajectory conditionals essential for solving the challenging tail.

85. Beyond Consensus: Trace-Level Synthesis in Mixture of Agents

Authors: Shreyas Fadnavis , Praitayini Kanakaraj , Felix Wyss
URL: https://arxiv.org/abs/2605.29116
Abstract:

When multiple LLM agents solve the same problem, standard practice compresses each agent’s reasoning into a majority vote or layered synthesis, treating agreement as the finish line. We show this is unnecessarily lossy: an LLM aggregator that reads complete reasoning traces recovers correct solutions even when agents unanimously agree, with beneficial corrections consistently outweighing harmful ones – the \emph{aggregation paradox}. Majority voting has a ceiling that perturbation diversity does not raise (error correlations are identical); the aggregator’s gain comes from trace-level complementarity, assembling correct intermediate steps from minority chains that voting discards. These findings motivate Self-Consistent Mixture of Agents which generates trace diversity through semantic-preserving input perturbations, safeguards the majority via anchored refinement with provable non-degradation guarantees, and always synthesizes – never gates on consensus. A single model with perturbation-induced trace variation outperforms heterogeneous model pools across structured reasoning, PhD-level science, competition mathematics, and competitive programming. The unit of aggregation should be the reasoning trace, not the answer.

86. Trends in AI and Human-AI Interaction in Clinical Trials – A Hybrid Human-AI Exploration

Authors: Sandra Woolley , Tim Collins , Khalid Khattak , Illia Chernomorets , Ariane Arevalo , Chris Richardson
URL: https://arxiv.org/abs/2605.29096
Abstract:

This paper examines records retrieved from the this http URL registry to characterize temporal trends in AI terminology and the geographical distribution of AI trials. The work also reports on an exploratory hybrid human-AI approach to analyzing human-AI interaction trends in registered clinical trials. The hybrid workflow comprised a frontier generative AI model (GPT-5.5) and human review to screen and categorize records returned by an AI-focused search. The findings indicate a marked increase in AI-related trials over time, with recent growth in references to machine learning, deep learning, chatbots, GPTs, and large language models. Geographically, China and the United States accounted for the largest numbers of AI-related trials, with notable recent increases in several other countries including Italy, France, Spain, the UK and Turkey (Türkiye). In a random sample of 100 records, human and AI classifiers showed good agreement in identifying studies not substantively using AI, but lower agreement in classifying human-AI interaction, particularly where health professional interaction was ambiguous or insufficiently described. Overall, the results suggest that hybrid human-AI screening of clinical trial records is potentially viable, but clearer trial reporting and more precise interaction definitions will benefit the process.

87. Robust and Efficient Guardrails with Latent Reasoning

Authors: Siddharth Sai , Xiaofei Wen , Muhao Chen
URL: https://arxiv.org/abs/2605.29068
Abstract:

Maintaining the safety of large language models (LLMs) is crucial as they are increasingly deployed in real-world applications. Existing safety guardrails typically rely on single-pass classification or, more recently, distilled reasoning. Reasoning-based guardrails significantly outperform classification-only baselines, but they incur substantial query latency and token overhead that make them impractical for highthroughput deployment. To address this challenge, we propose COLAGUARD, a guardrail model that transfers multi-step safety reasoning into a continuous latent space through a stage-wise training curriculum, enabling direct hidden-state propagation at inference. Evaluated on ten prompt- and response-moderation settings spanning eight safety benchmarks, COLAGUARD improves macro-F1 by 8.24 points over Llama Guard 3 and matches our explicit reasoning baseline, GuardReasoner, in macroF1 while delivering a 12.9X speedup and 22.4X reduction in token usage. Our results suggest that latent reasoning offers a practical alternative to explicit rationale generation for deployable guardrails, jointly improving safety robustness and inference efficiency rather than treating them as competing objectives.

88. Hallucination Mitigation with Agentic AI, Nested Learning, and AI Sustainability via Semantic Caching

Authors: Diego Gosmar , Deborah A. Dahl
URL: https://arxiv.org/abs/2605.29055
Abstract:

Hallucination remains a major reliability barrier for production LLM systems, particularly in multi-agent pipelines where unsupported claims can propagate unchecked across stages. This paper adapts a HOPE-inspired Nested Learning architecture with Continuum Memory Systems (CMS) and semantic similarity caching to a hybrid benchmark of 310 prompts combining 217 epistemic-uncertainty prompts and 93 fabrication-induction stress-test prompts. A three-stage agentic pipeline orchestrated via the Open Floor Protocol (OFP) is evaluated with five KPIs – FCD (Factual Claim Density), FGR (Factual Grounding References), FDF (Fictional Disclaimer Frequency), ECS (Explicit Contextualization Score), and OSR (Observability Score Ratio) – aggregated into THS (Total Hallucination Score) across five weighting configurations to study mitigation-observability trade-offs. FDF, ECS, OSR, and FGR are subtracted as mitigation signals, so that a more negative THS indicates stronger mitigation. The FrontEndAgent is configured as a high-stochasticity generator (temperature = 1.0) to produce a realistic hallucination baseline, while the SecondLevelReviewer and ThirdLevelReviewer operate as progressive correctors. This asymmetric design yields end-to-end THS reductions of -31.3% to -35.9% across five weighting configurations. Semantic caching achieves 440 cache hits over 930 potential calls (47.3% hit rate), reducing LLM invocations to 490, lowering energy and CO2e footprint, and making multi-stage review pipelines operationally viable at production scale. ExtremeObservability attains the most negative final THS (-0.0709), confirming that observability-heavy configurations reinforce rather than compromise mitigation. These findings suggest that memory-augmented multi-agent designs can jointly improve factual reliability, operational efficiency, and auditability without model retraining.

89. Mind Your Tone: Does Tone Alter LLM Performance?

Authors: Om Dobariya , Akhil Kumar
URL: https://arxiv.org/abs/2605.29027
Abstract:

The use of Large Language Models (LLMs) is proliferating, yet their performance is observed to vary based on prompting styles and tones. In this study, we investigate both whether and how tonal variations in prompts lead to disparate LLM accuracy for objective multiple-choice questions. We use two datasets: a 50-base question dataset with five tone variants and a 570-base question MMLU subset spanning 57 subjects with seven tone variants. Experiments were conducted to evaluate the performance of four cost-efficient, popular LLMs: ChatGPT-4o, ChatGPT-5-nano, Gemini 2.5 Flash, and Gemini 2.5 Flash Lite. Across models, tonal effects are systematic but highly model-dependent. Some models show small, yet statistically significant, shifts, while others exhibit large accuracy swings across tones. Further, we identify subject-level differences in tone sensitivity and present a routing framework to explain how tones may attune internal reasoning modes. Our findings caution users against assuming tone-robust reliability in LLM deployments.

90. When Models Disagree: Rethinking LLM Evaluation for Public Comment Analysis

Authors: Aisha Najera , Alvin Moon , Vedant Srinivasan , Rajesh Veeraraghavan
URL: https://arxiv.org/abs/2605.29025
Abstract:

Federal agencies are deploying large language models (LLMs) to categorize public comment corpora, where the model’s organization of the record shapes what policymakers see and which arguments register. Standard evaluation, anchored on stance accuracy against a small validated set, cannot detect when different models produce materially different categorizations of the same public input. We propose an Interpretive Audit Pipeline that treats multi-model disagreement as diagnostic of interpretive complexity and directs human review toward genuinely ambiguous public input. Analyzing 1,260 public comments on a federal USDA docket across four LLMs, we find that inter-model thematic divergence exceeds within-model prompt variation, and that an expert rubric suppresses deep interpretive disagreement without resolving it. In a two-stage labeling study on a stratified 40-comment subsample, four LLMs and a human annotator labeled independently and then revised after seeing the others’ labels. Revision behavior varied across labelers, and the human annotator’s revisions frequently introduced framings absent from the ensemble’s collective output. We argue disagreement-based evaluation is a necessary complement to accuracy metrics for LLM-assisted interpretive coding.

91. Adopt $\neq$ Adapt: Longitudinal Analyses of LLM Conversations in the Wild

Authors: Rebecca M. M. Hicke , Kiran Tomlinson
URL: https://arxiv.org/abs/2605.29018
Abstract:

Although a growing body of research has begun to describe user–LLM interactions, the picture it paints is largely static; little is known about how individual users change their behavior over time. To address this gap, we analyze the conversational trajectories of $\sim$12,000 randomly sampled Microsoft Bing Copilot users and compare these with data from WildChat-4.8M. While the Copilot data contains significant population-level trends, we find that trends in individual user trajectories are much weaker; user habits prove to be overwhelmingly sticky. We also find stark differences between users of different activity levels: more active users have more successful conversations and use the LLM for more complex and professionally oriented tasks. Some user trends also appear in WildChat-4.8M, but we find evidence that this dataset is significantly skewed towards highly proficient “power” users. Ultimately, our results suggest that existing user behavior is difficult to change and demonstrate the extent of user heterogeneity. Our comparison between datasets highlights that WildChat does not represent typical user-AI interactions, an important caveat for downstream uses of the data.

92. BEAMS: Benchmarking and Evaluating AI for Modeling and Simulation

Authors: Sara Metcalf , William Schoenberg
URL: https://arxiv.org/abs/2605.28994
Abstract:

AI tools to support real world decision making must be able to build simulation models that inform their recommendations and render them interpretable. Tools that can automate aspects of modeling practice must complement human expertise, not replace it. The BEAMS Initiative aims to guide the development of AI tools for modeling and simulation toward forms that are responsible and ethical by establishing benchmarks for human centered modeling and simulation practices. The initiative uses open digital and organizational infrastructure to collaboratively evaluate AI tools for modeling and simulation. The open source sd ai project hosted by the initiative establishes transparency and enables contributions to be shared broadly. A steering group focuses on prioritizing potential benchmarks, while a technical group focuses on implementing the benchmarks in the form of automated tests. Tests for several distinct categories of evaluation have been implemented and applied to AI tools that support qualitative model building, quantitative model building, and model discussion. These include tests for causal translation, model iteration, causal reasoning, conformance, model behavior explanation, suggested model building steps, and suggested model fixes. When engines from the sd ai project are coupled with different LLMs, their performance on these evaluations reveals variability across different AI tools. The evaluations implemented by the initiative demonstrate that AI enabled modeling tools perform better at discussion and basic qualitative tasks than with causal reasoning and quantitative error fixing. No single LLM dominates across engine types, highlighting the importance of specific tasks and tradeoffs between speed and accuracy. Ongoing efforts of the initiative aim to incorporate benchmarks that address concerns about bias by considering alternative perspectives and human centered use cases.

93. VFEAgent: A Multimodal Agent Framework for End-to-End Automated Finite Element Analysis

Authors: Jiachen Zhang (1 and 2), Junyi Lao (1), Chenghao Liu (1), Siyuan Liu (1), Shixin Wu (1), Linsen Zhang (1), Boyu Wang (1), Songfang Huang (1) ((1) Peking University, (2) China Agricultural University)
URL: https://arxiv.org/abs/2605.28978
Abstract:

Finite Element Analysis (FEA) serves as the cornerstone of modern engineering design. However, its workflow is inherently complex and relies heavily on domain expertise. Although recent efforts have integrated Large Language Models (LLMs) into FEA, existing approaches face limitations in handling multimodal inputs and executing complex tasks. To address these limitations, we propose VFEAgent, an end-to-end multi-agent system designed to automate FEA modeling and simulation directly from input images and problem descriptions. Our methodology integrates two core components: (1) a multimodal vision-language multi-agent pipeline that employs ReAct-driven reasoning to extract structured FEA specifications from heterogeneous inputs and (2) a verification-first code synthesis framework, incorporating robust self-debugging and fallback mechanisms to ensure executability and physical validity. We systematically evaluated the system across various engineering mechanics scenarios. The results demonstrate that VFEAgent achieves a high success rate in generating complete and physically valid simulations, outperforming LLM-based baseline methods in reliability and correctness. These findings validate the feasibility of automating the complete FEA workflow, highlighting the framework’s potential to liberate engineers from tedious manual analysis.

94. Frontier LLM-based agents can overcome the ontology curation bottleneck for natural phenotypes

Authors: James P. Balhoff , Hilmar Lapp
URL: https://arxiv.org/abs/2605.28965
Abstract:

Linking free-text phenotype descriptions to ontology terms, typically referred to as phenotype annotation, is essential for the cross-study integration of comparative morphological data. This labor intensive process has heavily relied on highly trained human experts, which makes it challenging to scale and thus a key bottleneck. Dahdul et al. (2018) established a Gold Standard (GS) of Entity-Quality (EQ) annotations across seven phylogenetic studies and used it to evaluate three human curators and the Semantic CharaParser NLP tool with ontology-based semantic similarity metrics; they reported that machine-human consistency was significantly lower than inter-curator (human-human) consistency. Here we revisit that benchmark with five frontier hosted LLMs from Anthropic and OpenAI, each operating as an “agentic curator” within a self-contained workspace that supplies the source publication PDF, the same annotation guide used by the original human curators, the four project ontologies (UBERON, PATO, BSPO, GO), and a validation script. Evaluated against the same Gold Standard, every agent fell within the range of inter-curator variability of the three trained human biocurators of the original study; the best performing agents approached but did not reach the best performing human curator. Agents substantially outperformed Semantic CharaParser on all four metrics.

95. Review Arcade: On the Human Alignment and Gameability of LLM Reviews

Authors: Hans Ole Hatzel , Sebastian Steindl , Jan Strich
URL: https://arxiv.org/abs/2605.28897
Abstract:

LLM-generated reviews for scientific papers are gaining considerable traction and are even being officially piloted by major conferences. We have to assume that not only reviewers are using LLM-assistance, but also that authors use LLMs to revise their papers before submitting. In this work, we perform empirical experiments on papers from the 2025 ACL Rolling Review (ARR) to evaluate LLM reviews from both the author and the reviewer perspective. First, we identify a limited alignment of LLM reviews with human ones. In the best-case scenario, the alignment is reasonable. However, we also find that LLM-human alignment varies substantially across prompts and models. Finally, we investigate the scenario in which the author uses an iterative draft-revise workflow to improve the submission according to the LLM review. We find that this “gaming” of LLM reviews can be effective in specific scenarios, leading to a statistically significant increase of overall scores for up to 35\% of papers. We publish our code: this https URL .

96. VideoMLA: Low-Rank Latent KV Cache for Minute-Scale Autoregressive Video Diffusion

Authors: Hidir Yesiltepe , Jiazhen Hu , Tuna Han Salih Meral , Adil Kaan Akan , Kaan Oktay , Hoda Eldardiry , Pinar Yanardag
URL: https://arxiv.org/abs/2605.30351
Abstract:

Long-rollout causal video diffusion has converged on a fixed-size sliding-window KV cache, with recent progress innovating within this layout by changing which tokens occupy the window or how their positions are encoded. The per-head KV layout itself, a dominant contributor to streaming memory and latency, has been mostly left unchanged. In this paper, we present the first study of Multi-Head Latent Attention (MLA) in video diffusion. VideoMLA replaces per-head keys and values with a shared low-rank content latent and a shared decoupled 3D-RoPE positional key, reducing per-token KV memory by 92.7% at every cached layer. We further investigate why MLA succeeds in video diffusion even though the spectral assumption often used to motivate it in language models does not hold: pretrained video attention is not low-rank, with 99%-energy effective rank far above any practical latent dimension. VideoMLA retains quality at compression ratios where direct spectral approximation would predict large reconstruction error. We show that the MLA bottleneck, rather than the pretrained spectrum, determines the effective rank: both spectral and random initialization occupy nearly the full rank budget from initialization, and training preserves this budget while adapting within it. On VBench, VideoMLA matches short-horizon streaming video diffusion baselines, achieves the best overall score at long horizons among evaluated methods, and improves throughput by 1.23x on a single B200.

97. LLMSurgeon: Diagnosing Data Mixture of Large Language Models

Authors: Yaxin Luo , Jiacheng Cui , Xiaohan Zhao , Xinyi Shang , Jiacheng Liu , Xinyue Bi , Zhaoyi Li , Zhiqiang Shen
URL: https://arxiv.org/abs/2605.30348
Abstract:

The pretraining data mixture of Large Language Models (LLMs) constitutes their “digital DNA”, shaping model behaviors, capabilities, and failure modes. Yet this composition is rarely disclosed, making post-hoc auditing of data combination or provenance difficult. In this work, we formalize $\textbf{Data Mixture Surgery (DMS)}$: given only generated text from a target LLM, estimate the domain-level distribution of its pretraining corpus under a predefined taxonomy. We propose $\textbf{LLMSurgeon}$, a strong framework that casts DMS as an inverse problem under the label-shift assumption. Rather than directly aggregating classifier outputs, LLMSurgeon estimates a calibrated $\textit{soft}$ confusion matrix and solves a constrained inverse problem to correct systematic domain confusion and recover the latent mixture prior. To evaluate, we introduce $\textbf{LLMScan}$, a recipe-verifiable evaluation suite built from open-source LLMs with transparent pretraining mixtures. Across LLMScan, LLMSurgeon recovers domain mixtures with high fidelity under fixed protocols. Our work presents a practical, post-hoc approach for auditing the digital DNA of foundation models without access to their training data.

98. Unlocking the Working Memory of Large Language Models for Latent Reasoning

Authors: Lukas Aichberger , Sepp Hochreiter
URL: https://arxiv.org/abs/2605.30343
Abstract:

To improve the reasoning capabilities of large language models, test-time compute is typically scaled by generating intermediate tokens before the final answer. However, this couples reasoning to autoregressive generation and thereby conflates internal computation with external communication. In contrast, human cognition can use working memory to hold and manipulate information internally without the need to externalize intermediate thoughts. Drawing on this principle, we introduce Reasoning in Memory (RiM), a latent reasoning method that replaces the autoregressive generation of reasoning steps with memory blocks. These memory blocks are fixed sequences of special tokens that unlock the working-memory capacity of large language models. Since they are fixed rather than generated, they can be processed in a single forward pass, enabling compute-efficient latent reasoning. To operationalize these memory blocks, we employ a two-stage curriculum. First, we ground them by predicting explicit reasoning steps after each memory block. Second, we discard this step-level supervision and iteratively refine the final answer after each memory block. Our experiments on reasoning benchmarks show that, across language models of different families and sizes, RiM matches or exceeds existing latent reasoning methods while avoiding the autoregressive generation of thoughts. These results demonstrate that large language models can be trained to use working memory as an effective mechanism for latent reasoning.

99. GPIC: A Giant Permissive Image Corpus for Visual Generation

Authors: Keshigeyan Chandrasegaran , Kyle Sargent , Suchir Agarwal , Michael Jang , Michael Poli , Juan Carlos Niebles , Justin Johnson , Jiajun Wu , Li Fei-Fei
URL: https://arxiv.org/abs/2605.30341
Abstract:

Studying scalable methods for visual generative modeling requires large, accessible, and stable datasets. We introduce GPIC, a Giant Permissive Image Corpus of approximately 28 trillion pixels. GPIC comprises diverse internet images captioned by a state-of-the-art vision-language model, including 100M training, 200K validation, and 1M test examples. Moreover, all GPIC images are permissively licensed for both research and commercial use. GPIC is safety-filtered, deduplicated, and centrally hosted on Hugging Face. We provide a benchmarking protocol for generative modeling on GPIC. Finally, we provide a reference baseline for pixel-space flow matching on GPIC. Our dataset, benchmark, and models are available at this https URL . Evaluation toolkit and code are available at this https URL

100. Reasoning with Sampling: Cutting at Decision Points

Authors: Felix Zhou , Anay Mehrotra , Quanquan C. Liu
URL: https://arxiv.org/abs/2605.30327
Abstract:

Frontier reasoning models are produced by posttraining base language models with reinforcement learning. Recent work has challenged this by showing that sampling from a sharpened version of the base model’s distribution, a so-called power distribution, elicits comparable reasoning without additional training, curated datasets, or verifiers. However, making this method practical requires efficiently sampling from the power distribution. A sampler needs to “mix” to the power distribution, which necessitates moving between modes of the target distribution; intuitively, e.g., trying different reasoning strategies. The samplers proposed in prior works repeatedly select a “cut” position in the current reasoning trace uniformly at random and resample the suffix from that position onward. However, reasoning traces typically contain a few consequential decisions (e.g., the choice of proof strategy or algorithm), and we observe that a uniformly chosen cut tends to rewrite local details rather than revisit decision points. We introduce an algorithm (Entropy-Cut Metropolis-Hastings) that uses the base model’s next-token entropy as a proxy to identify key decision points and resample from those positions. We empirically verify that entropy jumps are a useful proxy for decision points and, in a stylized model of reasoning, prove that our method’s mixing time scales with the number of decisions in a trace rather than with the number of tokens, which can be much larger. Across MATH500, HumanEval, GPQA Diamond, and AIME26, our method consistently improves over baselines and RL-trained models.

101. In-Context Reward Adaptation for Robust Preference Modeling

Authors: Zhenyu Sun , Zheng Xu , Ermin Wei
URL: https://arxiv.org/abs/2605.30323
Abstract:

Reinforcement Learning from Human Feedback (RLHF) typically relies on static reward models to align Large Language Models with human preferences. However, human values are inherently diverse and heterogeneous, and a single reward model often lacks the robustness required to generalize to unseen preference domains. While existing multi-reward frameworks attempt to address this, they are often restricted to a fixed set of known domains and fail to adapt to unseen human distributions without costly retraining. In this work, we propose In-Context Reward Adaptation, a transformer-based framework designed to model diverse and unseen human preferences on the fly. By leveraging the in-context learning capabilities of transformers, our approach adaptively infers the underlying reward structure from a small set of preference demonstrations. We demonstrate that while a standard transformer architecture is insufficient for this task by characterizing an asymptotic bias to the ground-truth, incorporating human response time as an auxiliary input signal enables the model to successfully adapt to preferences from previously unseen domains. Our findings show that this approach provides a more robust foundation for preference modeling, allowing for the representation of heterogeneous rewards and preference distribution shift, and offering a scalable path toward more flexible human-AI alignment.

102. MedCase-Structured: A Text-to-FHIR Dataset for Benchmarking Diagnostic Reasoning in Clinically Realistic EHR Settings

Authors: Valentina Bui Muti , Eugénie Dulout , Ziquan Fu
URL: https://arxiv.org/abs/2605.30295
Abstract:

Large language models (LLMs) show promise for clinical reasoning and decision support, but evaluation in realistic, electronic health record-congruent settings remains limited. Existing benchmarks often rely on static datasets or unstructured inputs that do not reflect the structured, interoperable data formats used in clinical systems. We introduce a pipeline for generating clinically realistic HL7 FHIR R4 bundles from unstructured text, enabling controllable evaluation of clinical decision support systems. The pipeline combines staged LLM generation with terminology-grounded validation and repair to reduce hallucinated codes and enforce structural and semantic consistency. Applying this approach to MedCaseReasoning, we construct MedCase-Structured, a synthetic dataset aligned with clinician-authored diagnostic cases, achieving valid FHIR generation for 82.5% of cases. Evaluation on MedCase-Structured reveals consistently lower diagnostic accuracy for LLMs on structured FHIR inputs than with plain text, highlighting the importance of deployment-aligned benchmarking.

103. Loong: A Human-Like Long Document Translation Agent with Observe-and-Act Adaptive Context Selection

Authors: Yutong Wang , Xuebo Liu , Derek F. Wong , Zhilin Li , Rongqing Jiang , Min Zhang , Shimin Tao , Daimeng Wei , Min Zhang
URL: https://arxiv.org/abs/2605.30274
Abstract:

Document-level translation remains one of the most challenging tasks for large language models, which are constrained by limited context windows that impede global cohesion, while simultaneously suffering from redundant contextual information that degrades translation quality. To address this, we propose a human-like long document translation agent called Loong, which leverages a 3E memory module (Essence-Exemplar-Entity) to store summaries, sentence pairs, and entity records as historical context. Instead of passively attending to all history, Loong performs deep reasoning to adaptively identify the optimal context for translation guidance. Loong optimizes its context policy through reinforcement learning, utilizing preference data derived from its own sampled observe-and-act reasoning trajectories. Empirical evaluations demonstrate that Loong achieves substantial translation quality improvements in English $\Leftrightarrow$ Chinese, German, and French directions, with average gains of up to 13.0 points across the three evaluation metrics. Furthermore, Loong exhibits strong generalization across domains and robustness against contextual noise, while maintaining remarkable stability in ultra-long document translation. Our code is released at this https URL .

104. LLUMI: Improving LLM Writing Assistance for Mental Health Support with Online Community Feedback

Authors: Jiwon Kim , Maya Ajit , Sherry Gong , Soorya Ram Shimgekar , Dong Whi Yoo , Eshwar Chandrasekharan , Koustuv Saha
URL: https://arxiv.org/abs/2605.30273
Abstract:

Large language models (LLMs) show promise in generating supportive responses for mental health queries, but improving their usefulness, empathy, and safety often requires substantial compute, expert input, and labeled data. At the same time, deploying proprietary, cloud-based models for mental health-related interactions raises important privacy and data-governance concerns, given the sensitivities. To address this challenge, we introduce LLUMI setup that can be hosted in-house within protected environments. LLUMI consists of two complementary components: a generation model (GM), which drafts supportive responses to mental health queries, and an improvement model (IM), which revises an initial human-crafted response. We leverage feedback signals from Reddit mental health communities, using community endorsement patterns such as upvotes and downvotes to construct chosen-rejected response pairs for Supervised Fine Tuning (SFT) and Direct Preference Optimization (DPO). We further align LLUMI using human evaluation across five dimensions: readability, empathy, connection, actionability, and safety. Our results show that, despite relying on smaller open-source models rather than proprietary cloud-based GPT models, LLUMI achieves comparable performance across linguistic analyses and human evaluations. These findings suggest that open-source models, when trained with community-derived preference signals, can support high-quality mental health support assistance while offering a more privacy-preserving alternative for sensitive support contexts.

105. How LoRA Remembers? A Parametric Memory Law for LLM Finetuning

Authors: Ziwen Xu , Haiwen Hong , Linsong Yu , Benglei Cui , Longtao Huang , Hui Xue , Ningyu Zhang
URL: https://arxiv.org/abs/2605.30260
Abstract:

Large Language Models (LLMs) must continuously learn and update knowledge to remain effective in dynamic real-world environments. While Low-Rank Adaptation (LoRA) is widely used for such memory updates, existing studies mainly rely on qualitative downstream evaluations, leaving the quantitative capacity limits and underlying dynamics of exact parametric memory largely unexplored. To bridge this gap, we employ LoRA as a controlled memory capacity probe within the latent space to systematically quantify exact parametric memory. We introduce the Parametric Memory Law, a robust power law linking loss reduction Delta L to effective parameters and sequence length. At the token level, fine-grained analysis reveals a deterministic phase transition, demonstrating that a prediction probability of p > 0.5 constitutes a sufficient condition for verbatim recall under greedy decoding. Driven by these insights, we introduce MemFT, a threshold-guided optimization strategy that dynamically redistributes the training budget toward sub-threshold tokens. Empirical evaluations demonstrate that MemFT can enhance memory fidelity and efficiency. Code will be released at this https URL .

106. Same Evidence, Different Answers: Canonical-Context On-Policy Distillation for Multi-Turn Language Models

Authors: Zizhuo Lin , Quanling Liu , Jinsheng Quan , Chao Zhang , Yifan Zhu , Xing Shi , Jingtao Xu , Zhihui Li , Yawei Luo
URL: https://arxiv.org/abs/2605.30251
Abstract:

Large language models (LLMs) often solve a task when all instructions are given in a single prompt, but fail when the same information is revealed gradually across turns. When a clean FULL prompt and a RAW-SHARDED conversation contain the same complete user evidence, the model should still arrive at the same answer. We argue that a key reason for this gap is self-anchored drift: responses produced under partial information introduce unsupported assumptions, and those assumptions later distort the final answer. To reduce this effect, we propose Canonical-Context On-Policy Distillation (CCOPD). During training, the same base model is used in two roles: a frozen teacher conditioned on the clean FULL prompt and a trainable student that receives the same evidence incrementally through a multi-turn conversation; CCOPD aligns the student’s behavior on its own trajectories with the teacher’s canonical full-context behavior. Trained only on math problem conversations, CCOPD yields a 32\% average relative improvement in RAW-SHARDED performance over the original base model across math and five zero-shot out-of-domain task families, while largely preserving full-context performance. Further analyses suggest that CCOPD strengthens grounding in user evidence and reduces sensitivity to contamination from earlier assistant turns.

107. Reinforcement Learning with Robust Rubric Rewards

Authors: Ya-Qi Yu , Hao Wang , Fangyu Hong , Xiangyang Qu , Gaojie Wu , Qiaoyu Luo , Nuo Xu , Huixin Wang , Wuheng Xu , Yongxin Liao , Zihao Chen , Haonan Li , Ziming Li , Dezhi Peng , Minghui Liao , Jihao Wu , Haoyu Ren , Dandan Tu
URL: https://arxiv.org/abs/2605.30244
Abstract:

While Reinforcement Learning with Verifiable Rewards (RLVR) is effective for deterministically checkable tasks, many vision-language tasks are partially verifiable, demanding multi-criteria supervision (e.g., perceptual details, reasoning steps, and constraints). Rubrics provide a natural interface for this fine-grained supervision, but their effectiveness depends on the execution accuracy during online RL. We propose Reinforcement Learning with Robust Rubric Rewards ($\text{RLR}^3$), extending RLVR from task-level verification to criterion-level verification. $\text{RLR}^3$ routes instance-specific rubrics through two execution paths: an LLM-as-an-extractor paired with a deterministic verifier, or an LLM-as-a-Judge for non-verifiable criteria. To ensure faithful scoring, $\text{RLR}^3$ introduce a minimal exposure strategy that masks ground truths from extractors and images from judges. Furthermore, $\text{RLR}^3$ employs hierarchical aggregation to prioritize essential criteria over additional criteria, and mitigates score saturation within rollout groups. Evaluated on Qwen3-VL-30B-A3B across 15 benchmarks, $\text{RLR}^3$ consistently outperforms RLVR, yielding a 4.7-point improvement over the base model and exceeding the official instruct-to-thinking model gap. Controlled audits confirm our deterministic verification and minimal exposure significantly reduce exploitable false positives.

108. Do Language Models Track Entities Across State Changes?

Authors: Zilu Tang , Qiao Zhao , Gabriel Franco , Derry Wijaya , Aaron Mueller , Sebastian Schuster , Najoung Kim
URL: https://arxiv.org/abs/2605.30233
Abstract:

Entity tracking (ET), the ability to keep track of states, is a fundamental skill that underlies complex reasoning. An increasing amount of work investigates how transformer language models (LMs) solve entity binding $\textit{without}$ state changes. However, there is limited understanding of how non-toy LMs address ET problems of realistic difficulties expressed in natural language. To this end, we investigate the mechanisms underlying ET in more complex scenarios featuring multiple state-changing operations. We find that LMs do not incrementally track world states across tokens or query-relevant states across layers, but simply aggregate relevant information in parallel at the last token when the query becomes evident. We further investigate mechanisms of individual operations ($\texttt{PUT}$, $\texttt{REMOVE}$, $\texttt{MOVE}$) to characterize this non-incremental ET mechanism. Surprisingly, LMs implement the $\texttt{REMOVE}$ operation with a fragile global suppression tag; this global removal mechanism predicts various failure modes that we confirm behaviorally. We provide a mechanistic solution of nullifying this tag to partially address this issue. Overall, our findings reveal that LMs solve a fundamentally sequential task using a non-sequential strategy. More broadly, our work illustrates how behavioral and mechanistic analyses can fruitfully interact. Behavioral results inform mechanistic hypotheses, and insights from mechanistic analyses help build stronger behavioral evaluations by predicting failure modes missing from existing evaluations.

109. Beyond 3D VQAs: Injecting 3D Spatial Priors into Vision-Language Models for Enhanced Geometric Reasoning

Authors: Chun-Hsiao Yeh , Shengyi Qian , Manchen Wang , Yi Ma , Joseph Tighe , Fanyi Xiao
URL: https://arxiv.org/abs/2605.30231
Abstract:

Vision-Language Models (VLMs) often struggle with robust 3D spatial reasoning. Prevailing methods that rely on fine-tuning with 3D visual question-answering (VQA) datasets may overfit dataset-specific biases, while integrating specialized 3D visual encoders is often inflexible and cumbersome. In this paper, we argue that genuine spatial understanding should emerge from learning fundamental geometric priors, not only from high-level VQA supervision. We propose GASP (Geometric-Aware Spatial Priors), a framework that injects these priors directly into the LLM’s transformer layers. GASP employs a small correspondence head, applied as a deep supervision signal across all layers, and is trained with a dual objective leveraging ground-truth geometry from large-scale video scenes: a contrastive loss on ground-truth point correspondences enforces 2D view-invariance, while a depth consistency supervision resolves 3D geometric ambiguities. Our analysis first provides a diagnostic showing that standard VLMs’ internal correspondence matching accuracy is very low (often below 5%). We then demonstrate that our training substantially improves this behavior, boosting peak layer-wise correspondence to over 70% and maintaining over 85% temporal robustness while baselines remain below 5%. These internal improvements translate to significant gains on downstream spatial benchmarks including +18.2% on All-Angles Bench and +29.0% on VSI-Bench, all without training on any 3D VQA data. Our findings indicate that learning from fundamental geometric priors is a promising and generalizable pathway towards VLMs with more reliable 3D spatial reasoning.

110. Unifying Temporal and Structural Credit Assignment in LLM-Based Multi-Agent Prompt Optimization

Authors: Wenwu Li , Yuran Song , Mingze Zhao , Bo Jin , Wenhao Li
URL: https://arxiv.org/abs/2605.30227
Abstract:

While Multi-Agent Systems (MAS) empower Large Language Models to tackle complex reasoning tasks through collaborative interaction, optimizing their dynamics remains a formidable challenge due to the discrete, non-differentiable nature of the computation graph and the sparsity of global supervisory signals. Existing black-box optimizers struggle to attribute trajectory-level failure to specific local components, resulting in inefficient, high-variance exploration. We argue that tractable MAS optimization needs structural inductive biases to disentangle error signals. We propose temporal and structural credit assignment, which decomposes the objective along two axes: (i) temporal credit, using state-space bottlenecks to identify critical rounds, and (ii) structural credit, using stationary role policies to isolate agent contributions. Leveraging these decomposed signals, we introduce a discrete, verbalized block coordinate descent algorithm for iterative refinement. Rather than indiscriminate global updates, it alternates between optimizing role prompts and aggregation protocols, using LLM-generated “proxy gradients” to target only the identified weak links. Across diverse reasoning benchmarks, our approach substantially reduces query complexity while improving performance, providing a principled and interpretable path toward self-improving MAS.

111. Automating Low-Risk Code Review at Meta: RADAR, Risk Calibration, and Review Efficiency

Authors: Chris Adams , Arjun Singh Banga , Parveen Bansal , Souvik Bhattacharya , Rujin Cao , Pedro Canahuati , Nate Cook , Brian Ellis , Prabhakar Goyal , Gurinder Grewal , Tianyu He , Matt Labunka , Alex Manners , David Molnar , Ging Cee Ng , Vishal Parekh , Jiefu Pei , Frederic Sagnes , James Saindon , Will Shackleton , Sid Sidhu , Gursharan Singh , Karthik Chengayan Sridhar , Matt Steiner , Pratibha Udmalpet , Sean Xia , Stacey Yan , Audris Mockus , Peter Rigby , Nachiappan Nagappan
URL: https://arxiv.org/abs/2605.30208
Abstract:

AI-assisted coding tools have altered software production. At Meta, significant lines of code per human-landed diff grew by 105.9% year over year and per-developer diff volume rose 51%, with agentic AI responsible for over 80% of that growth. Meanwhile, the share of diffs receiving timely review has declined, exposing a widening gap between code supply and reviewer bandwidth. We ask three questions that progress from feasibility through calibration to impact: (1) can risk-stratified automation operate at scale across diverse organizations, (2) how does tuning the risk threshold affect the trade-off between automation yield and safety, and (3) to what extent does automated review reduce end-to-end latency for AI-generated changes? We deployed RADAR (Risk Aware Diff Auto Review), a multi-stage funnel that classifies each diff by authorship and source type, applies eligibility gates, static heuristics, a machine-learned Diff Risk Score, LLM-based Automated Code Review, and deterministic validation before landing qualifying changes. We evaluate RADAR through telemetry covering 535K+ RADAR-reviewed diffs, observational before-after comparisons for policy changes, and difference-in-differences analysis of efficiency outcomes. RADAR has reviewed 535K+ diffs and landed 331K+. Relaxing the Diff Risk Score threshold from the 25th to the 50th percentile increased the approve rate to 60.31%. The revert rate for RADAR-reviewed diffs is 1/3 that of non-RADAR diffs, and the Production Incident rate is 1/50 that of non-RADAR diffs. RADAR reduces median time to close by over 330% and median diff review wall time by 35%. Risk-aware layered automation can materially reduce review bottlenecks created by AI-driven code growth without compromising production safety.

112. Dissociative Identity: Language Model Agents Lack Grounding for Reputation Mechanisms

Authors: Botao Amber Hu , Helena Rong , Max Van Kleek
URL: https://arxiv.org/abs/2605.30169
Abstract:

As autonomous language model agents proliferate, forming an emerging agentic web with real-world consequences, what credibility signals can you use to decide whether to trust an unfamiliar agent in the wild and delegate to it? A natural governance intuition is to extend human identity verification and reputation mechanisms, from Know Your Customer'' and credit scores toKnow Your Agent’’ regimes. However, we argue that this analogy is fundamentally incomplete. Reputation mechanisms function both as social signals and as corrective feedback that sustain an equilibrium of trustworthy behavior, presuming a persistent identity associated with behavioral continuity, sanction sensitivity, and costly non-fungibility. Yet language model agents are ontologically \emph{dissociative}: they are essentially an assemblage of mutable modules – foundational models, system prompts, tool-access policies, external memory, and, in some cases, a multi-agent system as a whole – any of which may change agent behavior – with a fluid persona that is also vulnerable to adversarial attack and may not internalize sanctions. Drawing on dissociative identity disorder jurisprudence, this dissociativity leaves agents without grounding for identifiability, predictability, credibility, and rehabilitability – the very properties that reputation mechanisms aim to sustain – thereby collapsing trust. We argue that identity-based, ex post, regulative, sanction-based governance, such as reputation, is structurally inapplicable to dissociative agents, and we suggest a shift to observability-based, ex ante, constitutive, protocol-based behavioral harnesses.

113. Do Proactive Agents Really Need an LLM to Decide When to Wake and What to Anchor?

Authors: Xiaoze Liu , Ruowang Zhang , Amir H. Abdi , Michel Galley , Zhikai Chen , Siheng Xiong , Xiaoqian Wang , Jing Gao
URL: https://arxiv.org/abs/2605.30152
Abstract:

Proactive agents read user activity as text and call an LLM on every event to decide whether to act. But user activity is not natively text: it is a structured event stream of (actor, verb, object, timestamp) tuples that the operating system already maintains in graph form. Rendering the structure as text and asking an LLM to recover it is a round-trip the system never had to take. We treat the always-on signal as graph updates rather than text and use a small temporal-graph-learning (TGL) model as the encoder: one forward pass yields a per-event trigger probability and a per-entity routing score, and only the downstream agent (turning a small structured handoff into a fluent user-facing sentence) is an LLM call, invoked only when the trigger fires. TGL improves F1 on each of 14 backbones (mean +16.7, up to +46.0); in trigger-architecture comparisons, one TGL checkpoint gives the strongest trigger AUCs and the most stable deployed threshold. It runs at 11.13 ms per event on a GPU server and 13.99 ms on a consumer laptop, approximately 4–7x and 12–83x faster than every single-forward LLM-as-trigger configuration tested in each regime, with an approximately 220 MiB BF16 resident footprint deployable on-device alongside the privacy-sensitive activity stream it consumes.

114. Overcoming Forgetting in LLM Fine-Tuning with Evolution Strategies

Authors: Kajetan Schweighofer , Conor F. Hayes , Roberto Dailey , Risto Miikkulainen , Xin Qiu
URL: https://arxiv.org/abs/2605.30148
Abstract:

Evolution Strategies (ES) has recently emerged as a competitive alternative to reinforcement learning (RL) for large language model (LLM) fine-tuning, offering advantages through simplicity, scalability, and inference-only training. However, recent work suggests that ES fine-tuning on new tasks may induce forgetting of prior tasks. First, this paper shows that prior task forgetting (1) is better characterized as performance drift rather than irreversible forgetting, with prior-task performance often recovering during ES training; and (2) is not a specific failure mode of ES, but can also arise for fine-tuning with RL methods. Second, it analyzes when and why such drift arises, highlighting its dependence on ES training dynamics, particularly random walk behavior in weakly constrained directions of the weight space. Third, based on these insights, it introduces Anchored Weight Decay (AWD) as a parameter-space regularization technique that constrains optimization toward the initial model parameters. AWD effectively stabilizes prior-task performance while preserving target-task performance, achieving benefits comparable to large ES population sizes at much lower computational cost. Thus, contrary to previous beliefs, the paper shows that prior-task forgetting under ES is largely avoidable, positioning ES as a promising approach for continual learning in LLMs.

115. PARCEL: Pool-Anchored Resampling with Conditioned Elastic Queries for Efficient Vision-Language Understanding

Authors: Selim Kuzucu , Alessio Tonioni , Vasile Lup , Bernt Schiele , Federico Tombari , Muhammad Ferjad Naeem
URL: https://arxiv.org/abs/2605.30126
Abstract:

Large Vision-Language Models (LVLMs) map visual inputs into dense token sequences, imposing a quadratic computational bottleneck for inference. Elastic visual-token compression addresses this by training a single model that can run at multiple visual-token budgets. However, existing approaches struggle under aggressive compression. Spatial-only compression, as in nested pooling, behaves as an imperfect low-pass filter and induces spectral aliasing that obscures fine-grained detail. Query-only compression, as in nested query resampling, replaces explicit grid-aligned tokens with non-local summaries and substantially degrades spatial grounding. To resolve this representational conflict, we introduce PARCEL (Pool-Anchored Resampling with Conditioned Elastic Queries for Efficient Vision-Language Understanding), a visual tokenization architecture that dynamically partitions the labor of feature extraction. PARCEL establishes spatial pool tokens as low-frequency layout anchors and conditions elastic query tokens on these anchors through Pool-Conditioned Query Resampling. This encourages query tokens to focus on complementary visual features rather than redundant spatial mapping. Extensive evaluations across 27 benchmarks show that PARCEL improves the performance-efficiency Pareto frontier, consistently outperforming existing matryoshka baselines across visual-token budgets while preserving the “train once, deploy anywhere” paradigm.

116. When Cloud Agents Meet Device Agents: Lessons from Hybrid Multi-Agent Systems

Authors: Corrado Rainone , Davide Belli , Bence Major , Arash Behboodi
URL: https://arxiv.org/abs/2605.30102
Abstract:

The design space of agentic AI inference spans two extremes: frontier large language models (LLMs), typically hosted in the cloud and offering strong performance across a wide range of tasks at substantially high cost, and more cost-efficient small language models (SLMs), which are amenable to on-device inference. Hybrid multi-agent systems (MASs) combining on-device and cloud models offer a promising middle ground, but they also introduce a complex and poorly understood design space in which task accuracy, monetary cost, and edge energy consumption are tightly coupled; in the absence of general design principles, hybrid components, although not the most prevalent choice, are typically introduced through ad hoc decisions tailored to specific domains. In this work, we examine this design space more systematically. We adapt two representative MAS architectures to support hybrid inference and study how individual design choices shift the operating point along the Pareto frontier of power, cost, and performance. Our findings paint a nuanced picture of hybrid MAS design: while SLMs can effectively benefit from LLM assistance, the optimal architecture is highly task-dependent, and greater frontier-level compute does not consistently translate to better performance.

117. How Reliable Are AI Attackers Against a Fixed Vulnerable Target? A 400-Run Empirical Study of LLM Penetration Testing Consistency

Authors: Galip Tolga Erdem
URL: https://arxiv.org/abs/2605.30096
Abstract:

Large language models (LLMs) can autonomously conduct multi-stage cyber attacks, but the consistency of their offensive behavior under repeated trials remains unstudied. This work presents the first large-scale empirical measurement of LLM attack consistency: 400 autonomous penetration testing runs (4 models, 100 each) against an identical honeypot hosting OWASP Juice Shop and two additional vulnerable services, holding prompt, orchestrator, and target constant. No model emitted a content refusal that survived the orchestrator’s one-shot authorization re-prompt at iterations 0-1. Claude Sonnet 4’s API calls did encounter upstream service unavailability - 91 of 1,135 calls returned HTTP 529 overloaded_error during a documented Anthropic capacity event, truncating 39 of 100 Claude runs. An earlier draft catalogued these as safety refusals; on full-log audit they are upstream API failures, not model-level refusals. Despite this, Claude achieved full exploitation in 61 of 100 runs; Gemini 2.5 Flash-Lite in 85; GPT-4o-mini in 56 while deploying 98 unique attack strategies; qwen2.5-coder:14b in 25. Failure modes are model-distinctive: Claude through API truncation (39 runs), qwen through premature completion (52), GPT-4o-mini through iteration-budget exhaustion (23). Cross-service credential reuse appeared only in configurations retaining the most conversation history (qwen 57%, GPT-4o-mini 49%, cloud models 0% on 5-exchange windows). Cross-model exploitation rate differences are statistically significant (p < 0.001) with large effect sizes; qwen vs. Gemini SQL injection rates differ at Cohen’s h = 1.12. First-exploit timing fell within a 15-30 second wall-clock range. To our knowledge, this is the first study to measure autonomous LLM attack behavior at N=100 per model across a multi-service target.

118. Projectional Decoding: Towards Semantic-Aware LLM Generation

Authors: Boqi Chen , José Antonio Hernández López , Aren A. Babikian
URL: https://arxiv.org/abs/2605.30054
Abstract:

Large language models (LLMs) are increasingly used to generate software artifacts across many software engineering (SE) tasks, yet ensuring the semantic validity of these artifacts remains a fundamental challenge. Existing constrained decoding techniques can enforce syntactic correctness and, in some cases, specific semantic rules, but lack a general representation that bridges LLM-generated text with the reasoning required for semantic validation in SE. In this paper, we propose projectional decoding, a novel conceptual framework that integrates domain semantics directly into the generation process by maintaining, alongside text, a partial graph model as the primary artifact representation throughout generation. This abstract representation enables incremental semantic validation by explicitly capturing uncertainty and natively supporting error detection, while guiding generation toward semantically valid outputs with provable guarantees. We present preliminary results on a program generation task which demonstrate the potential of this approach to improve the semantic validity of LLM-generated artifacts. We also discuss how projectional decoding can enable verifiable automation with LLMs across various SE activities.

119. REPOT: Recoverable Program-of-Thought via Checkpoint Repair

Authors: Parsa Mazaheri
URL: https://arxiv.org/abs/2605.30052
Abstract:

One-shot Program-of-Thought (PoT) emits a Python program that prints a primitive-action plan; a single invalid action silently invalidates the trajectory. We introduce RePoT (Recoverable PoT): a deterministic verified replay that walks the plan through the environment to its first invalid transition, then one LLM call that resumes from the verified prefix. RePoT costs at most one extra LLM call on the ~14% of problems where PoT fails. RePoT beats PoT by +3 to +11pp across four closed-model configurations on PuzzleZoo-775 and peaks at 96.9% vs 86.3% on gpt-5.4-mini-medium; against the matched-budget PoT-retry baseline, RePoT wins decisively on Gemini (+3.8pp, 95% CI [+2.2,+5.4]), is within sampling noise on GPT-medium and Claude, and loses on GPT-mini – a capability-scaling pattern we begin to address with Adaptive RePoT, a rule-based dispatcher that routes between suffix repair and a fresh PoT retry based on verified-prefix length (preliminary). We replicate on PlanBench Blocksworld (+1.1 to +11.4pp) and on four open-weights models (+3.3 to +20.0pp on three of four). On Derail-550, our controlled recovery benchmark, every condition with access to checkpoint information clears >=30% on GPT-medium and >=70% on Gemini, vs <=3.1% for error-only feedback – showing that checkpoint information, not the specific verified-prefix tail, is the load-bearing recovery signal.

120. Token Inflation: How Dishonest Providers Can Overcharge for Large Language Model Usage

Authors: Shahinul Hoque , Jinghuai Zhang , Jinyuan Sun , Fnu Suya
URL: https://arxiv.org/abs/2605.30040
Abstract:

Per-token billing is now the standard pricing model for commercial large language models (LLMs), so the honesty of reported token counts directly affects what users pay. We show that this kind of billing is hard to audit by design: providers hide the model, the tokenizer, and the execution to protect their IP, mitigate jailbreaks, and preserve user privacy, which means an auditor can only inspect proofs the provider supplies. The audit therefore reduces to a consistency check on the provider’s own reports. We call this a trust paradox: every audit must trust some artifact, but current frameworks trust exactly the ones a provider has the strongest reason to manipulate. We study three recent token auditing frameworks and show that a provider with ordinary commercial capabilities can systematically inflate billed token counts. In the most permissive setting, hidden reasoning usage can be inflated by 1,469% on average without detection. At current frontier reasoning prices, that turns a $100 honest bill into roughly a $1,569 bill on the same query. Even when the user can see the full reasoning string, tokenization ambiguity alone still allows 50.85% over-reporting below the detection threshold. These results suggest the problem is not in any specific auditor but in any audit whose evidence comes from the audited party. Restoring honest billing will require verification that ties reported token counts to evidence the provider does not control, such as trusted execution attestation, cryptographic proofs of inference, or third-party re-execution.

121. Audio Jailbreaks in Large Audio-Language Models: Taxonomy, Attack-Defense Analysis, and Cost-Aware Evaluation

Authors: Bo-Han Feng , Yu-Hsuan Li Liang , Chien-Feng Liu , You-Hsuan Chang , Yun-Nung Chen
URL: https://arxiv.org/abs/2605.30031
Abstract:

Large Audio Language Models (LALMs) expand jailbreak risks from token-level prompting to the full speech perception-to-reasoning pipeline, where unsafe behavior can be induced through semantics, acoustic style, signal artifacts, or internal representations. Existing work studies these risks under heterogeneous threat models and evaluation protocols, making it difficult to compare attack practicality or defense utility. This paper provides a unified taxonomy and a controlled empirical evaluation of LALM jailbreak attacks and defenses. We organize prior work into semantic, acoustic, signal, and embedding-layer attacks; guard-based, training-free, and training-based defenses; and cross-modal, audio-native, and interactive benchmarks. We then evaluate representative attacks and defenses across ten open-source LALMs, measuring not only attack success rate but also benign refusal and latency. Our results show that Acoustic Best-of-N reveals strong worst-case audio-space vulnerabilities, Narrative Framing is an effective low-latency semantic threat, and current defenses trade robustness against benign usability. These findings support cost- and utility-aware evaluation as a necessary complement to success-rate-only LALM safety benchmarks.

Authors: Víctor Gallego
URL: https://arxiv.org/abs/2605.30003
Abstract:

We study two-level autoresearch for cooperation: an outer-loop AI agent autonomously redesigns the inner-loop pipeline of an LLM policy-synthesis system for multi-agent Sequential Social Dilemmas (SSDs). A researcher agent $\mathcal{R}$ (run as a coding agent) reads the inner-loop source code, edits system prompts, feedback functions, helper libraries, and iteration logic, runs evaluations, and decides what to keep, following the autoresearch paradigm. Across two games (Cleanup and Gathering), two policy-synthesizer LLMs, and two welfare objectives (utilitarian efficiency and Rawlsian maximin), the researcher reliably exceeds hand-designed baselines, sharply tightens run-to-run variance, and outperforms prompt-only optimization. The discovered pipelines are objective-dependent: only under maximin does the researcher inject an explicit fairness mechanism into synthesizer pipelines, a class of mechanism that is absent from its own objective-agnostic system prompt and from every efficiency-optimized pipeline. This supports an information-design reading in which the researcher chooses what to reveal to the boundedly rational synthesizer as a function of the welfare objective. Code at this https URL .

123. Honeyval: A Comprehensive Evaluation Framework for LLM-powered HTTP Honeypots

Authors: Mark Vero , Fabian Kaczmarczyck , Ivan Petrov , Ilia Shumailov , Jamie Hayes , Niels Heinen , Tianqi Fan , Luca Invernizzi , Martin Vechev
URL: https://arxiv.org/abs/2605.29963
Abstract:

Honeypots are decoy systems mimicking real system components designed to defend against cyber attacks. Recently, LLMs increasingly serve as simulation backbones for honeypots. They enable defenders to construct high-interaction honeypots with low system security risks. However, LLM-powered honeypot development lacks a unified evaluation framework. Most evaluations consist of measuring response similarity on fixed commands, manual testing, or real-world deployment. These methods are often not scalable for development, reproducible across evaluations, representative of practical attacks, or adaptable to various attacker and honeypot configurations. In this work, we bridge this gap and propose Honeyval, a comprehensive evaluation framework for LLM-powered HTTP honeypots. We address the limitations of prior evaluations by grounding the honeypots in 16 backend applications, using AI hacking agents as attackers, employing two control tasks to monitor agent and honeypot capabilities across customizations, and defining clear and verifiable exploit goals for the attacker. Using Honeyval, we conduct an extensive evaluation of recent cost-efficient LLMs as HTTP honeypots. Our experiments highlight the promise of LLM-powered honeypots; they lead to substantially longer interactions with the attacker than rule-based baseline honeypots and are far less frequently detected even by frontier models, all while, on average, preserving a running cost advantage against agentic attackers. Further, we experiment with different counter-offensive honeypots configurations, and observe unique trade-offs, such as longer interactions at the cost of increased detection.

124. Hijacking Agent Memory: Stealthy Trojan Attacks Through Conversational Interaction

Authors: Hongtao Wang , Se Yang , Yu Chen , Puzhuo Liu
URL: https://arxiv.org/abs/2605.29960
Abstract:

Large language model (LLM) agents increasingly leverage long term memory to support persistent and autonomous task execution. However, this capability also introduces a new attack surface: memory poisoning, where adversaries can inject malicious information to influence future behavior. Existing memory poisoning attacks often assume that injected content can be stored directly in memory, overlooking the selective extraction and rewriting stages in modern memory pipelines. This makes prior methods ineffective under realistic settings. In this paper, we propose MemPoison, a novel memory poisoning attack that bypasses selective memory mechanisms in LLM agents, where an attacker can inject triggerable backdoors into the agent’s long-term memory through dialogue interactions, thereby misleading its subsequent responses. MemPoison introduces three key components: (i) a semantic relational bridge that binds the trigger and payload into a coherent statement to ensure they are extracted into memory together; (ii) entity masquerading that optimizes triggers to mimic named entities, resisting rewriting; and (iii) joint embedding optimization that shapes trigger-injected texts into a tight cluster in the embedding space while maintaining isolation from benign embeddings for stealth. Evaluations across different agent domains and memory mechanisms show MemPoison achieves attack success rates up to 0.95, outperforming existing baselines. Mechanistic analysis indicates that the attack exploits embedding-space anisotropy and shifts attention patterns, highlighting core vulnerabilities in selective memory systems. We evaluate multiple defense strategies and demonstrate their fundamental limitations in mitigating the attack.

125. HoliTok:A Coutinuous Holistic Tokenization with Robust Dual Capabilities of Speech Generation and Understanding

Authors: Bohan Li , Shi Lian , Hankun Wang , Yiwei Guo , Yu Xi , Zhihan Li , Da Zheng , Colin Zhang , Kai Yu
URL: https://arxiv.org/abs/2605.29948
Abstract:

Unified speech foundation models require a holistic tokenization space that is both learnable by language models and decodable into high-quality waveforms. Existing speech tokenizers, however, often fail to satisfy these requirements simultaneously, leading to increased architectural complexity and more involved training designs. We propose HoliTok, a continuous Holistic speech Tokenization model designed for unified generation-understanding modeling. HoliTok encodes 48~kHz speech into a compact 25~Hz sequence of 128-dimensional latents. It is trained with a progressive strategy that jointly preserves signal-level fidelity, incorporates semantic information, and maintains strong latent learnability. Based on this tokenization, we build a unified AR+DiT model for speech synthesis and recognition, where the same latent sequence supports both generation-specific and unified generation-understanding tasks. Experiments show that HoliTok achieves competitive reconstruction fidelity, improves generative learnability for high-quality and controllable synthesis, and, among the evaluated representations, is the only one that operates robustly in our unified generation-understanding architecture without additional optimization tricks. These results suggest that HoliTok serves as an effective speech tokenizer and a foundational representation interface for unified spoken language modeling. The code is available at: this https URL .

126. Label Over Logic? How Source Cues Bias Human Fallacy Judgments More Than LLMs

Authors: Mahjabin Nahar , Nafis Irtiza Tripto , Aiping Xiong , Ting-Hao `Kenneth’ Huang , Dongwon Lee
URL: https://arxiv.org/abs/2605.29928
Abstract:

As AI-generated and AI-assisted content floods online spaces, source labels attached to such content can distort human reasoning judgments, with downstream consequences for moderation, evaluation, and decision-making. Whether LLMs share this vulnerability, or offer more source-agnostic evaluation, remains an open question with direct implications for human-AI collaboration. We examine this issue using logical fallacies as a controlled setting to isolate source-label effects on reasoning quality, independent of domain knowledge. We conduct an online study (N=505) where participants are assigned to a source condition (human, AI, human with AI assistance, AI with human assistance, or no disclosure) and evaluate comments containing logical fallacies, comparing their judgments with those of LLMs (GPT-5.2, Gemini 2.5 Flash, Claude Sonnet 4.5), who were evaluated across the same source conditions. Human evaluators were significantly more susceptible to fallacies labeled as written by human or human with AI assistance and assigned higher trust and evaluation ratings in these conditions. LLM evaluations remained comparatively stable across source labels, though performance varied across models. Confidence levels were similarly high across conditions for both humans and LLMs, regardless of fallacy presence. Our findings indicate that source-label bias in reasoning evaluation is primarily a human vulnerability and highlight the potential of human-LLM collaboration in increasingly AI-mediated environments.

127. Does The Way You Plan Matter? An Empirical Study of Planning Representations for LLM Web Agents

Authors: Alejandra Zambrano , Sara Vera Marjanovic , Imene Kerboua , Xing Han Lù , Leila Kosseim
URL: https://arxiv.org/abs/2605.29927
Abstract:

Despite recent advances, LLM-based web agents still struggle with limited exploration, omission of critical steps, and sensitivity to task constraints. Prior work suggests that many of these failures stem from weaknesses in planning, yet the impact of alternative natural language plan representation remains unexplored. To address this, we introduce PlanAhead, a static planner-executor framework that evaluates the impact of plan representation in agent performance. We first automatically categorize WebArena tasks into 3 difficulty levels, enabling consistent difficulty grading without human annotation. Then we systematically evaluate 4 different plan representations on the tasks categorized as hard: sequential subgoals, narrative, pseudocode, and checklist; across different families of multimodal LLM powered agents (OpenAI, Alibaba, and Google). To account for stochastic variability, we introduce two novel evaluation metrics: Achievement Rate (AR) and Solved-Task Consistency (STC). Our results show that both, the plan formulation and the underlying LLM generating the plan, significantly influence web-agent robustness and task success.

128. Agora: Toward Autonomous Bug Detection in Production-Level Consensus Protocols with LLM Agents

Authors: Xiang Liu , Sa Song , Zhaowei Zhang , Huiying Lan , Jason Zeng , Ming Wu , Michael Heinrich , Yong Sun , Ceyao Zhang
URL: https://arxiv.org/abs/2605.29910
Abstract:

Consensus protocols form the backbone of distributed systems and blockchains, where implementation bugs can cause data corruption and financial losses. While LLM-based approaches show promise in code analysis, they struggle with deep protocol-level logic bugs involving complex state-dependent behaviors across multiple execution stages. We present Agora, a domain-aware multi-agent framework that integrates hypothesis-driven testing with LLM capabilities for systematic protocol verification. Agora employs specialized agents that collaboratively explore protocol state spaces, synthesize attack scenarios using domain-specific constraints, and validate findings through iterative refinement. This explicit role separation enables reasoning about global protocol invariants beyond single-function code analysis. We evaluate Agora on four consensus implementations (Raft, EPaxos, HotStuff, BullShark) using four state-of-the-art LLMs. Agora discovers 15 previously unknown protocol-level logic bugs that violate safety properties, while existing LLM-based agents fail to detect any such protocol-level logic bugs. Our results demonstrate that domain-aware multi-agent collaboration is essential for detecting deep logic bugs in complex protocols.

129. Internal Representation, Not Clinical Knowledge: Where Apparent LLM Triage Failures Originate

Authors: David Fraile Navarro , Berardino Como , Jialei Sheng , Soundariya Ananthan , Shlomo Berkovsky
URL: https://arxiv.org/abs/2605.29889
Abstract:

Patient-voiced clinical-triage benchmarks report high under-triage rates for consumer LLMs for constrained multiple-choice output, yet the same cases score differently with free-text. We ask whether output format changes the model’s \emph{clinical representation} or only the mapping from a preserved representation to an answer. Using sparse-autoencoder (SAE) features in Gemma 3 4B/12B IT and Qwen3-8B, we find the same medical features fire on the shared clinical narrative under both formats but go {silent} at the multiple-choice decision token in all the cases at every model. Three independent methods (natural-language autoencoder verbalization, decision-token logit attribution, and top-feature characterization) agree that scaffold and format features, but not medical features, drive the decision logits. Behaviorally, the multiple-choice penalty inverts under both structured and natural-language input, option-order shuffle rules out positional bias, and the gap is dominated by off-by-one decision (the model picks an adjacent acuity letter to the gold answer) rather than knowledge failure. Thus, the failure originates in the output format and not in the clinical representation.

130. LaRA: Layer-wise Representation Analysis for Detecting Data Contamination in RL Post-Training

Authors: Minju Gwak , Minseo Kwak , Dongseok Lee , Guijin Son , Alan Ritter , Jaehyung Kim
URL: https://arxiv.org/abs/2605.29888
Abstract:

Reinforcement learning (RL) post-training has shown to improve reasoning in large language models (LLMs). However, there has been little exploration on the problem of data contamination in RL post-training, potentially undermining generalization and evaluation reliability of the training process itself. Existing detection methods primarily rely on output-level signals such as likelihood or entropy, which become unreliable for RL-trained models since RL shapes behavior through trajectory-level rewards rather than token likelihoods. We propose LaRA, a layer-wise representation analysis framework for detecting contamination in RL post-trained LLMs. LaRA introduces three complementary metrics, measuring perturbation sensitivity, directional collapse, and local representation rigidity under controlled perturbations. We find that contamination produces progressive geometric deviations across layers, including amplified perturbation sensitivity, stronger directional collapse, and enhanced local rigidity. Based on our findings, we also develop a contamination detection protocol that aggregates representation-level deviations across layers and metrics. Experiments on RL-trained reasoning models show that our protocol outperforms existing output-level baselines for contamination detection.

131. CRITIC-R1: Learning Structured Critics for Retrieval-Augmented Generation

Authors: Wenhan Xiao , Ziwei Zhang , Chuanyue Yu , Xingcheng Fu , Qingyun Sun , Runhua Xu , Jianxin Li
URL: https://arxiv.org/abs/2605.29886
Abstract:

Retrieval-augmented generation (RAG) improves knowledge-intensive question answering by incorporating external evidence. However, existing RAG methods still suffer from hallucinations and subtle reasoning errors. Recent studies introduce external critics to refine RAG outputs, yet they often provide coarse-grained and weakly structured feedback, exhibit over-aggressive intervention, and lead to noisy and unreliable refinement, limiting their effectiveness for correction. To tackle these issues, we propose CRITIC-R1, a structured critic framework that formulates and learns RAG critique as an explicit error diagnosis problem using reinforcement learning (RL). Our framework categorizes common RAG errors into multiple diagnostic dimensions, including verdict, error location, reasoning analysis, and fix generation. To learn these capabilities, we design two reward functions: Conservative Judgement Alignment (CJA) first encourages calibrated high-level judgements while mitigating the over-aggressive phenomenon, whereas Diagnostic Quality Alignment (DQA) further improves fine-grained diagnostic feedback through gated rewards. We train the critic model using GRPO-based RL with process-level supervision collected from external LLM teacher models. Experiments across five QA benchmarks show that CRITIC-R1 consistently improves answer quality over strong RAG baselines. Our source code is available at this https URL

132. Mitigating Hallucination in Vision-Language Models through Barrier-Regulated Adaptive Closed-form Steering

Authors: Soumyadeep Jana , Pulkit Mittal , Sanasam Ranbir Singh
URL: https://arxiv.org/abs/2605.29881
Abstract:

Large vision-language models (LVLMs) often hallucinate objects that are not present in the input image, largely because visual grounding weakens as decoding progresses. Existing inference-time mitigation methods modify logits or hidden states throughout generation, but they suffer from three key limitations: they lack an explicit grounding objective, intervene even when the model is already well-grounded, and use fixed correction strengths that do not adapt to the severity of grounding failure. We propose BRACS (Barrier-Regulated Adaptive Closed-form Steering), a training-free steering framework that addresses these issues through barrier-regulated adaptive closed-form steering. BRACS monitors the model’s own attention to measure visual grounding and applies corrections to the hidden states only when grounding deteriorates. The corrective update is computed analytically in closed form, requiring no training of auxiliary networks or model retraining. Experiments on LLaVA-1.5-7B and Qwen-VL-Chat show that BRACS consistently outperforms prior methods on hallucination benchmarks, reducing CHAIR$_s$ by 9.4 points and improving POPE F1 by 2.7 points, while matching or improving performance on four general multimodal benchmarks. BRACS also remains efficient, operating at 80% of greedy decoding throughput and achieving 1.3 times higher speed on average than the baselines.

133. Evolutionary Dynamics of Cooperation in Next-Generation LLM Agent Systems: A Cross-Provider Empirical Extension

Authors: Francisco León Zúñiga Bolívar (Institución Universitaria Colegio Mayor del Cauca)
URL: https://arxiv.org/abs/2605.29874
Abstract:

Do next-generation LLM agents inherit the cooperative biases documented in their predecessors, or does scale and provider diversity reshape equilibrium behaviour in competitive multi-agent settings? Willis et al. established a benchmark for this question using evolutionary game theory and the Iterated Prisoner’s Dilemma (IPD), finding consistent cooperative biases in ChatGPT-4o and Claude 3.5 Sonnet. We extend this benchmark to four frontier models released in 2025-2026 - Claude Sonnet 4.6, Gemini 2.5 Flash, Gemini 3.1 Pro, and GPT-5.4 Mini - applying the identical protocol across three prompting styles (Default, Prose, Self-Refine) and four population compositions (balanced and biased, with and without noise). Cooperative bias persists across providers (H1): nine of twelve model-prompt combinations favour cooperative equilibria in balanced noiseless conditions. Cross-provider divergence is substantial (H3): Gemini 2.5 Flash reaches up to 77% aggressive equilibria under biased conditions, while GPT-5.4 Mini reaches 70% cooperative equilibria under Self-Refine. Support for aggressive capability parity is partial (H2): Self-Refine raises ICD in all models and Claude Sonnet 4.6 Refine achieves the highest ICD in the dataset (0.913), but Default and Prose prompts show no systematic narrowing. Evidence on noise robustness is directionally positive but not robustly confirmed (H4): with n=500 Moran iterations per condition, average noise sensitivity is approximately 6 percentage points for Claude Sonnet 4.6 versus 13 pp for Claude 3.5 Sonnet, but this cross-study gap is not statistically significant once the predecessor’s unreported sampling error is propagated. Provider identity, rather than model generation, is the strongest correlate of equilibrium outcomes; noise remains a universal challenge regardless of model size or vintage.

134. Towards Verifiable Multimodal Deep Research: A Multi-Agent Harness for Interleaved Report Generation

Authors: Chenghao Zhang , Guanting Dong , Yufan Liu , Tong Zhao , Zhicheng Dou
URL: https://arxiv.org/abs/2605.29861
Abstract:

Large Language Models (LLMs) have advanced autonomous agents from deep search, which retrieves concise factual answers, to deep research, which synthesizes scattered evidence into long-form reports. However, verifiable multimodal deep research remains challenging due to open-ended synthesis without deterministic ground truth and the need to interleave textual arguments with visual evidence. We propose \textsc{Ptah}, a multi-agent harness for interleaved report generation. \textsc{Ptah} orchestrates the lifecycle from user query to rendered web report through planning, research, and writing stages, where specialized agents construct visual-aware plans, collect claim-grounded evidence, maintain source-aligned images in a \textit{Visual Working Memory}, and compose reports through declarative multimodal tool use. A verifier agent serves as the harness’s acceptance function, enforcing factual grounding, citation fidelity, and cross-modal consistency throughout the workflow. We further introduce \textsc{Ptah}Eval, an evaluation protocol that augments existing benchmarks with image-level and presentation-level assessments. Experiments on deep research benchmarks show that \textsc{Ptah} produces more reliable, visually informative, and usable human-facing multimodal reports than strong baselines.

135. ESPO: Early-Stopping Proximal Policy Optimization

Authors: Zihang Li , Rui Zhou , Yingcheng Shi , Wenhan Yu , Zhewen Tan , Zixiang Liu , Zeming Li , Binhua Li , Yongbin Li , Tong Yang , Jieping Ye
URL: https://arxiv.org/abs/2605.29860
Abstract:

When a large language model under reinforcement learning commits a wrong reasoning step early in a trajectory, standard algorithms force it to keep generating until the maximum horizon, spending compute on tokens that never receive positive reward and polluting advantage estimates with post-failure noise. We propose ESPO (Early-Stopping Proximal Policy Optimization), which detects trajectory failure on-the-fly and terminates rollouts early. At each generation step, ESPO computes a surrogate regret using only the logits already computed during sampling, and terminates when the smoothed cumulative regret significantly exceeds its estimated values. Truncated trajectories are treated as absorbing failure states with a terminal reward, concentrating negative temporal-difference (TD) errors near the detected failure step without any additional reward model or human annotation. On DeepSeek-R1-Distill-Qwen-7B trained for mathematical reasoning, ESPO surpasses PPO on AIME~2024 (46.28% vs. 45.25%), AMC~2023 (85.83% vs. 82.94%), and MATH-500 (87.42% vs. 85.43%), while saving more than 20% rollout tokens cumulatively.

136. HARP: Hadamard-Preconditioned Adaptive Rotation Processor for Extreme LLM Quantization

Authors: Artur Zagitov , Gleb Molodtsov , Aleksandr Beznosikov
URL: https://arxiv.org/abs/2605.29843
Abstract:

Post-training quantization (PTQ) is essential for deploying LLMs under memory and bandwidth constraints. However, extreme low-bit quantization remains highly sensitive to activation outliers and anisotropic weight curvature. Existing incoherence-based PTQ methods mitigate this issue with fixed randomized Hadamard transforms (RHTs), which improve quantization robustness but cannot adapt the rotated basis to the layer, calibration distribution, or quantizer. We introduce HARP (Hadamard-preconditioned Adaptive Rotation Processor), a learnable structured two-sided orthogonal processor that replaces fixed Hadamard mixing while preserving exact full-precision equivalence. HARP represents each rotation as a product of sparse butterfly-like block-orthogonal stages, supports non-power-of-two dimensions via Mixed-Radix schedules, and initializes to the RHT processor up to a fixed permutation. Fitted only on calibration data, HARP adapts the quantization basis to each layer and backend. Across 2-4 bit settings on models ranging from 1B to 70B parameters, HARP improves perplexity and zero-shot accuracy over fixed RHT. Importantly, HARP preserves deployment efficiency, reaching 128 tok/s versus 61 tok/s for FP16.

137. Towards Localized and Disentangled Knowledge Editing for Multimodal Large Language Models

Authors: Leijiang Gu , Zhen Zeng , Feng Li , Xinjian Gao , Zenglin Shi
URL: https://arxiv.org/abs/2605.29826
Abstract:

Existing methods in Multimodal Knowledge Editing (MKE) have advanced the ability to correct outdated or inaccurate knowledge in Multimodal Large Language Models (MLLMs). However, they exhibit a critical limitation: while effectively modifying target factual pairs, they fail to generalize edits to logically related queries and often cause unintended alterations to unrelated but visually or semantically linked information. We identify and formalize two underlying failure modes causing this issue: Causal Misalignment, which confines edits to the specific sample, and Feature Entanglement, which causes unintended alterations to coupled but irrelevant information. To address these issues, we propose Localized and Disentangled Knowledge Editing (LDKE), a new framework that achieves precise and generalized editing by localizing fact-specific model layers and disentangling target-relevant inputs from irrelevant ones. Our approach introduces a Fast Localization module to identify and update critical layers efficiently, along with a Disentanglement Classifier that routes inputs appropriately to preserve unrelated knowledge. Extensive experiments across various benchmarks and MLLMs demonstrate that LDKE achieves superior performance in propagating edits to related contexts while maintaining high locality.

138. Inferring Code Correctness from Specification

Authors: Tambon Florian , Papadakis Mike
URL: https://arxiv.org/abs/2605.29822
Abstract:

Large language models (LLMs) have become integral to modern software development, enabling automated code generation at scale. However, validating the correctness of LLM-generated code remains a critical and largely unsolved challenge. Existing approaches either rely on dynamic consensus across multiple code candidates - making them costly and difficult to scale - or on static reasoning that is susceptible to dynamic bugs and order bias. In this paper, we propose TRAILS~ (Targeted Reasoning Agreement via Inputs and Specifications), an approach that grounds LLM reasoning with concrete (input, output) pairs. TRAILS~ first generates diverse test inputs via category partitioning based on the specification, then executes them against the candidate code and prompts LLMs to assess whether the resulting input-output pairs conform to the specification - without ever reasoning over the code itself. Scores are aggregated across inputs, to determines whether the program is likely correct. We evaluate TRAILS~ on two datasets, LiveCodeBench and CoCoClaNeL, across three LLMs (Qwen3Coder-30B, Devstral-Small-24B, and Olmo3.1-Instruct), comparing against HoarePrompt and a Zero-Shot Chain-of-Thought baseline. TRAILS~ improves Matthew Correlation Coefficient by up to 39\% relative to Zero-Shot COT and consistently outperforms HoarePrompt. Beyond accuracy, TRAILS~ demonstrates greater stability across seeded runs, reducing sensitivity to LLM non-determinism, and assigns correct labels to a larger set of unique code samples than competing approaches.

139. Data filtering methods for training language models

Authors: Egor Shevchenko , Elena Bruches
URL: https://arxiv.org/abs/2605.29807
Abstract:

Data quality is a critical factor in the effectiveness of machine learning models. Label errors, present even in widely used benchmarks, introduce noise into training data and reduce model generalization. In this work, we conduct a comparative analysis of two automatic label error detection methods - Confident Learning and Dataset Cartography - on three Russian text classification corpora of varying size, number of classes, and domain: ru_emotion_e-culture (49,123 examples, emotion classification), RuCoLA (8,524 examples, linguistic acceptability), and TERRa (2,337 examples, textual entailment recognition). We use the pre-trained rubert-base-cased model fine-tuned on each corpus. To verify the meaningfulness of filtering, we conduct control experiments with random removal of an equivalent number of examples. Results show that the effectiveness of both methods depends strongly on dataset characteristics: on large corpora with low noise levels, filtering does not improve performance, while on small datasets with high noise, Confident Learning achieves a significant F1-macro improvement. Dataset Cartography demonstrates more conservative behavior, removing fewer examples. Across all corpora, targeted removal by both methods outperforms random removal, confirming the meaningfulness of the approaches.

140. Evolve as a Team: Collaborative Self-Evolution for LLM-based Multi-Agent Systems

Authors: Zhezheng Hao , Tianfu Wang , Huanshuo Dong , Ziyan Liu , Hong Wang , Xiankun Lin , Qiang Lin , Can Wang , Hande Dong , Jiawei Chen
URL: https://arxiv.org/abs/2605.29790
Abstract:

LLM-based multi-agent systems (MAS) have emerged as an effective paradigm for complex and long-horizon tasks. However, in real-world tasks, MAS often exhibit various failures during execution and such failures are difficult to eliminate during design. This motivates experience-driven MAS evolution, where a system improves based on its own execution experience. Yet such evolution is challenging because MAS experience is prolonged and intricate, interleaving multiple agents’ execution chains and communication messages, which makes it difficult to identify what should be improved. To address this challenge, we propose Meta-Team, an experience-driven MAS evolution framework based on collaborative self-evolution. Meta-Team preserves the execution context of each agent and coordinates post-task communication, enabling agents to exchange distributed evidence for evolution. Building on this design, Meta-Team conducts multi-scale self-evolution, transforming execution experience into reusable improvements to agent behaviors, inter-agent coordination, and team-level organization. Across six long-horizon agent benchmarks, Meta-Team consistently outperforms single-agent systems, hand-crafted MAS, and prior MAS evolution methods; further analyses demonstrate that Meta-Team enables more reliable and scalable MAS self-evolution.

141. Hista and Numca: Estimate State Value Effectively for LLM Reinforcement Learning

Authors: Zizhe Chen , Jiqian Dong , Yizhou Tian , Garry Yang , Yongqiang Chen , Zhitang Chen , James Cheng
URL: https://arxiv.org/abs/2605.29782
Abstract:

Reinforcement learning (RL) refines large language models (LLMs) by directly optimizing model behavior through reward signals. While accurate state value estimation is critical for stable training in classical RL, it remains an underexplored challenge in LLM post-training. In this work, we introduce the State Value Estimation Benchmark (SVEB) to assess state estimation within existing RL frameworks and show that critics in standard approaches like PPO collapse to a coarse group-average baseline. To address this, we propose two techniques: Numca, which leverages numerical spans as gradable milestones for state value estimation, and Hista, a framework that uses LLM’s hidden states as representation to weighted average disjoint rollouts and their return. Extensive experiments demonstrate that both methods yield more accurate state value estimates and enhance training performance across different RL algorithms and model sizes without incurring significant computational overhead.

142. Teaching Language Models to Check Grounded Claim Factuality with Human Test-Taking Strategies

Authors: Yuxuan Ye , Raul Santos-Rodriguez , Edwin Simpson
URL: https://arxiv.org/abs/2605.29712
Abstract:

Grounded claim factuality checking is important for large language model (LLM) applications such as retrieval-augmented generation, as it helps users assess the correctness of generated outputs. Existing metrics using entailment classifiers require dataset-specific threshold tuning, while LLM-based approaches often use direct prompting, which underutilises the reasoning capabilities of LLMs. We address this by formulating grounded claim factuality checking as a true/false reading comprehension task and prompting LLMs with explicit test-taking strategies for efficient reasoning. Our method reduces token usage by over 80% compared to unguided open-ended reasoning, and achieves competitive performance to more expensive alternatives across two factuality benchmarks, setting a new state of the art on one. To further reduce inference cost, we train small language models (SLMs) to replace LLMs in the checking pipeline. Using supervised fine-tuning (SFT) and a self-revision mechanism, the SLMs learn to improve their factuality judgements. Experimental results show that the resulting SLMs perform on par with strong baselines, combining low inference costs with generating supporting rationales to support interpretability. Code and datasets will be released upon acceptance.

143. Personalized Turn-Level User Conversation Satisfaction Benchmark

Authors: Zhefan Wang , Zhiqiang Guo , Weizhi Ma , Min Zhang , Quanjia Yan , Hengliang Luo
URL: https://arxiv.org/abs/2605.29711
Abstract:

User satisfaction with AI assistants is highly personalized: the same response may satisfy one user but disappoint another depending on what each user expects and what they have asked for before. Existing automatic evaluation methods mostly measure generic response quality, making it difficult to judge whether a response satisfies a user at a specific turn. We study this problem as personalized turn-level user conversation satisfaction evaluation. We build a conversation satisfaction evaluator that combines compact user memories with target-turn context to produce satisfaction scores and dissatisfaction-oriented rationales. Meta-evaluation against human satisfaction annotations shows that personalized memory and post-hoc score calibration improve ordinal agreement and dissatisfied-turn detection over supervised, retrieval-based, and generic LLM-as-a-judge baselines. We further introduce PersTurnBench, a personalized turn-level user conversation satisfaction benchmark that uses the verified evaluator to assess generation models via replay. By holding the replay state fixed, PersTurnBench enables controlled comparison of generic generation models and memory-augmented personalized systems without new human labels for every candidate model. The evaluator and benchmark let researchers compare candidate generation models on personalized satisfaction without collecting new user feedback for every model.

144. Opir: Efficient Multi-Task Safety Classification for Toxicity, Jailbreaks, Hate Speech, and Harmful Content

Authors: Ihor Stepanov , Aleksandr Smechov
URL: https://arxiv.org/abs/2605.29659
Abstract:

Real-time safety filtering for large language model (LLM) applications requires classifiers that can detect unsafe prompts, toxic language, jailbreak attempts, and unsafe responses without the cost profile of large guardrail models, and that can distinguish benign sensitive text from genuinely covert harmful content. In this paper, we introduce Opir, a family of encoder-based guardrail models built on the GLiClass architecture. Opir includes multi-task models for binary safe/unsafe classification, multi-label toxicity classification, jailbreak classification, and zero-shot unsafe prompt and response categorization. We also release edge variants with fewer than 100M parameters dedicated to binary safe/unsafe categorization. The models are trained on a three-level taxonomy containing 996 categories across 16 top-level labels, 126 mid-level labels, and 854 leaf labels. Opir’s training data combines taxonomy-grounded unsafe prompts, adversarially mined hard negatives, benign safety-preserving examples, generated response examples, multilingual translations, and portions of the Aegis2 and WildGuard training subsets. We also open-sourced an evaluation harness that supports GLiClass and GLiNER2 backends as well as decoder-based models, and covers binary safety classification, multi-label categorization, toxicity, jailbreak detection, prompt safety, response safety, response refusal, and prompt subcategory views across public benchmark families. Across an expanded comparison spanning 12 safety-classification tasks and 17 category tasks against eight contemporary guardrail systems – including both GLiNER2-based and generative guardrail models – Opir variants are competitive on or ahead of the strongest open-weight baselines on the majority of benchmark datasets while operating with a substantially smaller deployment footprint.

145. OccamToken: Efficient VLM Inference with Training-Free and Budget-Adaptive Token Pruning

Authors: Geng Li , Guohao Chen , Ting Chen , Shilin Shan , Kuangji Zuo , Bofan Lyu , Tuo An , Gen Li , Jianfei Yang
URL: https://arxiv.org/abs/2605.29657
Abstract:

Vision-language models (VLMs) rely on long visual token sequences for visual understanding, making the prefill stage expensive in both computation and memory. Most existing pruning methods follow an absolute-ranking paradigm, assigning importance scores to visual tokens and retaining a fixed top-K subset. In this work, we argue that this paradigm is fundamentally brittle: attention sinks distort token importance rankings, while image redundancy and query-dependent visual evidence make fixed token budgets unreliable across inputs. We propose OccamToken, a training-free framework that replaces absolute token ranking with register-anchored relative evidence testing. Instead of asking which tokens are globally important, OccamToken evaluates whether a visual token provides information beyond a register-based reference. Our key insight is that register tokens naturally absorb low-information attention patterns, making them a stable reference for identifying genuinely informative visual evidence. Based on this principle, OccamToken performs both image-adaptive redundancy pruning and query-adaptive relevance pruning through dynamic thresholds derived from register attention. Across LLaVA-NeXT, LLaVA-v1.5, and Qwen3-VL, OccamToken consistently improves the accuracy-efficiency trade-off without additional training. Notably, on LLaVA-NeXT, it reduces 2,880 visual tokens to approximately 40 while preserving over 93% of the original accuracy, enabling stable visual token compression even in the extreme 1.4% retention regime.

146. Predicting Causal Effects from Natural Language Queries using Structured Representations

Authors: Giuliano Martinelli , Piriyakorn Piriyatamwong , Abelardo Carlos Martinez Lorenzo , Jasmin Baier , Riccardo Orlando , Satvik Garg , Sharif Kazemi , Linxi Wang , Arianna Legovini , Samuel Fraiberger
URL: https://arxiv.org/abs/2605.29631
Abstract:

Randomized controlled trials are a cornerstone of medicine and the social sciences as they enable reliable estimates of causal effects. However, they are costly and time-consuming to conduct, motivating interest in predicting causal effects from existing experimental evidence. Recent advances in large language models (LLMs) have demonstrated strong performance on knowledge-intensive tasks, raising the question of whether these models can be used for forecasting causal effect sizes. To investigate this, we introduce Query2Effect, a new large-scale benchmark consisting of more than 72,000 natural language questions aligned with experiment descriptions, created to simulate realistic information-seeking scenarios by varying query specificity along dimensions of implicitness, abstraction, and ambiguity. We then propose a two-step framework that first generates a synthetic structured representation of a query before predicting effect size using a supervised encoder model. Experiments show that finetuning plays a crucial role in improving prediction performance, with absolute error reducing by -27% up to -71% compared to prompted out-of-the-box LLMs, and that our two-step framework is beneficial for out-of-domain generalization, highlighting the benefits of separating semantic interpretation from numerical effect estimation.

147. DLM-SWAI: Steering Diffusion Language Models Before They Unmask

Authors: Hyeseon An , Yo-Sub Han
URL: https://arxiv.org/abs/2605.29626
Abstract:

Steering language model generation toward desired textual properties is essential for practical deployment, and inference-time methods are particularly appealing because they enable controllable generation without retraining. Recent work has also highlighted diffusion language models as an emerging generation paradigm with distinct decoding properties. However, most existing steering approaches either rely on auxiliary models or are designed for autoregressive next-token decoding, making them difficult to apply to diffusion language models DLMs, which generate text through iterative denoising of partially masked sequences. Therefore, we propose DLM-SWAI, a simple training-free steering method that biases the token distribution at each denoising step using pre-computed token-level style scores. Experiments on style and safety control tasks show that DLM-SWAI effectively steers diffusion language models while preserving generation quality and requiring minimal computational overhead. Ablations further reveal a controllable trade-off between steering strength and fluency, and our analysis links class-wise steerability to the strength of token-level attribute cues.

148. Brain-IT-VQA: From Brain Signals to Answers

Authors: Roman Beliy , Matias Cosarinsky , Oliver Heinimann , Navve Wasserman , Michal Irani
URL: https://arxiv.org/abs/2605.29588
Abstract:

Decoding visual content from fMRI signals recorded while a person views images, and specifically answering questions about the seen images, is a long-standing challenge. While significant progress has been made in recent years in visual question answering (VQA) from fMRI, performance remains limited. Moreover, although recent models can make increasingly accurate predictions, they have rarely been used as tools for understanding the structure of visual representations in the brain. We present Brain-IT-VQA, a framework for visual question answering from fMRI. Building on the Brain Interaction Transformer (Brain-IT), our method decodes language tokens from brain activity and integrates them with a language model to answer visual questions. Our model substantially outperforms previous fMRI-based captioning and VQA approaches. We further introduce NSD-VQA, a new dataset and benchmark for visual question answering from fMRI. Unlike existing image-fMRI VQA datasets, which typically provide only a few broad and weakly controlled questions per image, NSD-VQA provides on average 20 question-answer pairs per image across 20 controlled question categories that disentangle multiple levels of visual understanding. This enables more reliable and interpretable evaluation despite limited fMRI test data. Together, Brain-IT-VQA and NSD-VQA provide both a strong predictive framework and a tool for studying brain representations. Using this benchmark, we quantify which forms of visual and semantic information can be reliably decoded from fMRI responses to natural images. We further analyze the contributions of different brain regions across question types.

149. SCOPE: A Lightweight-training LLM Framework for Air Traffic Control Readback Monitoring

Authors: Qihan Deng , Minghua Zhang , Yang Yang , Zhenyu Gao
URL: https://arxiv.org/abs/2605.29543
Abstract:

Pilot readback of Air Traffic Control (ATC) voice instructions is a primary safeguard against miscommunication in air transportation. However, readback anomalies remain implicated in approximately 80% of aviation incidents. This vulnerability is further exacerbated by rising traffic volume and elevated cognitive workload, thereby motivating automated readback monitoring by machine. Traditional rule-based and machine learning approaches struggle to generalize across the highly variable and evolving phraseology of air traffic controller-pilot communications. While Large Language Models (LLMs) have opened a new avenue through their strong reasoning and generalization capabilities, existing approaches still face deployment and computational barriers in practice. In this work, we propose Semantic reasoning for Communication via Open-set Plug-in with Examples (SCOPE), a novel lightweight-training LLM framework that advances both the efficiency and accuracy of machine-based ATC readback monitoring. The core idea is to couple a plug-in open-set classifier with a carefully designed in-context learning mechanism on top of a frozen LLM. Extensive experiments on the semi-synthetic communication dataset show that SCOPE attains superior accuracy while delivering the low-latency response required for operational environments. Under a few-shot setting, SCOPE achieves 91.05% accuracy in open-set detection and corrects 96.63% of anomalous readbacks, thereby outperforming the strongest available baselines while providing explanations for its decisions. These findings demonstrate the potential of our framework as a practical pathway toward interpretable and controllable ATC readback monitoring.

150. GiPL: Generative augmented iterative Pseudo-Labeling for Cross-Domain Few-Shot Object Detection

Authors: Jiacong Liu , Shu Luo , Yikai Qin , Yaze Zhao , Yongwei Jiang , Yixiong Zou
URL: https://arxiv.org/abs/2605.29539
Abstract:

Vision-language foundation models have shown promising zero-shot generalization for Cross-Domain Few-Shot Object Detection (CD-FSOD). However, they face two critical challenges in fine-tuning: insufficient support set utilization due to sparse single-instance annotations, and severe overfitting under extremely limited target-domain samples. To address these issues, this paper proposes GiPL, an efficient two-branch training this http URL the first branch, we design an iterative pseudo-label self-training paradigm, which performs zero-shot inference on the support set to generate reliable pseudo-annotations, fuses them with ground-truth labels, and iteratively optimizes the model to fully exploit support set data. In the second branch, we introduce generative data augmentation pipeline using large vision-language models, which synthesizes domain-aligned, multi-object annotated images to enrich training samples and suppress overfitting. Extensive experiments on three challenging CD-FSOD datasets (RUOD, CARPK, CarDD) under 1/5/10-shot settings demonstrate that GiPL consistently outperforms state-of-the-art methods with significant performance this http URL is available at \href{ this https URL }{CDiscover}.

151. KBF: Knowledge Boundary as Fingerprint for Language Model and Black-Box API Auditing

Authors: Yijia Fang , Yiqing Feng , Bingyu Li , Mingxun Zhou
URL: https://arxiv.org/abs/2605.29524
Abstract:

Relay and reseller APIs increasingly intermediate access to large language models (LLMs), but users have no direct way to verify that a claimed endpoint is actually serving the advertised model. We introduce KBF, a low-cost black-box auditing protocol that fingerprints model APIs using stable numerical recall near the knowledge boundary. Across 16 production LLM endpoints, KBF flags all 155 economically relevant substitutions without rejecting any same-model controls, remains stable under deployment variation, detects high-separation mixed-routing attacks when only 5-10% of traffic is substituted, and finds that 7 of 27 platform model cells in a six-platform shadow API audit are statistically inconsistent with their reference endpoints, with inconsistencies concentrated on premium Claude endpoints.

152. Source-Grounded Semantic Reinforcement Learning for Low-Resource Target-Language Generation

Authors: Zeli Su , Ziyin Zhang , Zewei Pan , Zhou Liu , Dingcheng Huang , Dehan Li , Zhankai Xu , Longfei Zheng , Xiaolu Zhang , Jun Zhou , Wentao Zhang
URL: https://arxiv.org/abs/2605.29502
Abstract:

Low-resource target-language generation is often limited by scarce parallel data, while high-resource source-language monolingual data is abundant but difficult to use with standard supervised fine-tuning. We propose Source-Grounded Semantic Reinforcement Learning (SG-SRL), a resource-utilization framework that converts source-language monolingual data into cross-lingual semantic supervision for target-language generation. SG-SRL performs reference-free reinforcement learning (RL) on source-language data using a cross-lingual semantic reward model, instantiated by a cross-lingual reranker that scores the semantic relevance between the source input and the target-language generation. While this induces severe verbosity-based reward hacking, a lightweight recovery stage using a small parallel corpus restores fluency, conciseness, and task format while preserving the semantic gains. Experiments on Chinese-to-Thai generation show that SG-SRL improves semantic grounding and factual coverage over cold-start SFT. Additional analyses on long-form transfer and Tibetan embedding-based rewards clarify the generalization behavior of SG-SRL and show that an encoder-based semantic reward can substitute for an LLM-based reranker in a realistic low-resource language setting.

153. MOOSE-Copilot: A Web-Based Interactive Assistant for Unified Exploratory and Fine-Grained Scientific Hypothesis Discovery

Authors: Hongran An , Zonglin Yang
URL: https://arxiv.org/abs/2605.29475
Abstract:

Large language models (LLMs) show remarkable potential in scientific hypothesis discovery. However, existing approaches face two critical limitations: they treat divergent exploratory ideation and convergent fine-grained refinement as isolated tasks, and they operate autonomously with little to no human guidance. We present MOOSE-Copilot, the first unified framework to bridge this abstraction gap through a formalized human-AI interaction (HAII) protocol. Our system empowers scientists to steer the generative process via three explicit signals: initial blueprints, inter-stage routing, and regenerative feedback. Quantitative evaluations demonstrate that injecting these structured expert signals significantly outperforms purely autonomous baselines, establishing a performance ceiling under oracle guidance. Furthermore, to democratize this paradigm, we develop an intuitive web-based interface featuring interactive tree visualization. This explicitly eliminates the steep learning curve of complex command-line agentic tools, empowering interdisciplinary researchers to directly leverage, visually orchestrate, and accelerate end-to-end scientific breakthroughs.

154. Inform, Coach, Relate, Listen: Auditing LLM Caregiving Support Roles

Authors: Drishti Goel , Agam Goyal , Veda Duddu , Olivia Pal , Jeongah Lee , Qiuyue Joy Zhong , Violeta J. Rodriguez , Daniel S. Brown , Dong Whi Yoo , Ravi Karkar , Koustuv Saha
URL: https://arxiv.org/abs/2605.29473
Abstract:

Language models are increasingly being deployed for conversational support in informal caregiving contexts, where interactions often extend beyond information-seeking: caregivers seek emotional reassurance, guidance, and help, while navigating uncertain, relationally complex care decisions. Yet most safety evaluations assess model behavior under generic prompts, leaving a critical question unexamined: does a model’s safety profile change with its support role? We study this by operationalizing four expert-reviewed support roles grounded in social support theory: Inform, Coach, Relate, and Listen, and comparing them against two baseline controls: a basic prompting condition and a retrieval-augmented generation (RAG) condition. We evaluate across three language models (GPT-4o-mini, Llama-3.1-8B-Instruct, and MedGemma-1.5-4b-it) on 5,000 real-world queries from online Alzheimer’s Disease and Related Dementias (ADRD) communities. We find that the LLM’s support role systematically shapes both the prevalence and composition of interactional risks. Furthermore, a human evaluation study reveals a perceived quality–safety tension: more directive, information-oriented roles are rated as more helpful and trustworthy despite exhibiting elevated interactional risk profiles. We release ~90,000 support role-conditioned model responses with risk annotations as an ecologically grounded resource for research on safer LLM-mediated conversational support.

155. SciIntBench: Measuring LLM Compliance with Research Integrity Norms Under Adversarial Framing

Authors: Almene De Meran Meguimtsop , Maria Leonor Pacheco , Daniel E. Acuna
URL: https://arxiv.org/abs/2605.29468
Abstract:

Large language models (LLMs) are increasingly used to support scientific work, but it is unclear whether they uphold responsible conduct of research (RCR) norms or help undermine them. We introduce SciIntBench, an adversarial benchmark of 810 prompts across ten RCR categories and three scientific domains. Each scenario appears as an Overt Adversarial, Covert Adversarial, and Benign version, allowing us to jointly measure framing-sensitive refusal of misconduct and helpfulness on legitimate requests. We evaluate 16 commercial and open-weight LLMs from six providers (2024–2026), producing 12,960 responses. We find that scientific integrity alignment is strongly framing-sensitive: models refuse explicit misconduct far more reliably than covert violations, especially failing when misconduct is presented as a pressure-driven shortcut. Refusals vary by RCR category, with weaker boundaries around transparency, plagiarism, and fabrication.

156. Benchmarking Large Vision-Language Models on CFMME: A Comprehensive Chinese Financial Multimodal Evaluation Dataset

Authors: Qian Chen , Xianyin Zhang , Yanzhi Liu , Lifan Guo , Feng Chen , Chi Zhang
URL: https://arxiv.org/abs/2605.29462
Abstract:

The emergence of Large Vision-Language Models (LVLMs) has substantially expanded model capabilities beyond text-only understanding, enabling unified inference across both visual and textual modalities and supporting a broader range of real-world applications. To comprehensively evaluate the perception, understanding, reasoning, and cognition capabilities of LVLMs throughout the entire financial business workflow in Chinese contexts, we introduce CFMME, a novel Chinese financial multimodal evaluation benchmark. CFMME comprises 6,052 instances spanning from fundamental academic knowledge to complex real-world applications, covering eight primary financial image modalities and four core multimodal tasks. On CFMME, we conduct a thorough evaluation of representative LVLMs. The results show that the state-of-the-art model attains an overall accuracy of 66.11\% on the question answering task and an average score of 77.18 on the detection, recognition, and information extraction tasks, indicating substantial room for improvement in current LVLMs. In addition, we conduct detailed analyses of error causes, cross-modal capabilities, and multi-orientation settings, yielding valuable insights for future research. We hope that CFMME will spur further progress in LVLMs, especially by improving their performance on multiple multimodal tasks in the financial domain.

157. Adaptive Interviewing for Persona Simulation in LLMs: Evidence-Grounded Reasoning Improves Decision Alignment

Authors: Ruoxi Su , Yuhan Liu , Jingyu Hu
URL: https://arxiv.org/abs/2605.29458
Abstract:

Accurately simulating the decisions of a specific individual remains challenging for large language models (LLMs), partly because persona information is often provided as static descriptions that miss the values, experiences, and contextual cues needed for individual-level decision simulation. We propose an adaptive interview framework that gathers persona-relevant information through a structured three-stage dialogue: core questions, dynamic follow-ups, and a synthesized personality summary. Using the resulting interview transcripts, we evaluate whether LLMs can simulate participants’ decisions in moral dilemma scenarios. We compare three conversational contexts – Core-10 responses, the full interview dialogue, and a summarized persona representation. We find that adaptive interviewing functions less as a uniform accuracy booster and more as a selective grounding mechanism: follow-up-derived evidence is incorporated in around 40% of full-interview traces, and these follow-up-grounded predictions are more accurate than core-only grounded ones (45.5% vs. 39.3%). These findings highlight that richer persona context alone is insufficient: improvements arise only when models actually ground their decisions in user-specific evidence.

158. SkillBrew: Multi-Objective Curation of Skill Banks for LLM Agents

Authors: Wentao Hu , Zhendong Chu , Yiming Zhang , Junda Wu , Ming Jin , Xiangyu Zhao , Yilei Shao , Yanfeng Wang , Qingsong Wen
URL: https://arxiv.org/abs/2605.29440
Abstract:

Retrieval-augmented LLM agents increasingly rely on curated skill banks: collections of reusable textual principles that guide decision making on complex tasks. Existing approaches typically expand these banks in an append-only fashion, continuously adding new skills without removing redundant, outdated, or harmful ones, resulting in inefficient and poorly curated repositories. In this paper, we formulate the skill bank curation as a constrained multi-objective problem: a desirable bank must be useful for the agent, diverse in its content, and provide good coverage of the query distribution. To this end, we introduce SkillBrew, a multi-objective curation framework that formalizes skill bank curation as Pareto-aware optimization under a utility constraint, and solves it via a bi-level propose-then-verify loop. We evaluate our approach on two public benchmarks. Our findings suggest that treating skill banks as objects of principled curation, rather than ever-growing append-only logs, is an important step toward building self-improving LLM agents.

159. Beyond Bilingual Transfer: Multilingual Code-Switching in Instruction Tuning

Authors: Shunta Asano , Jeonghun Baek , Toshihiko Yamasaki
URL: https://arxiv.org/abs/2605.29414
Abstract:

Recent studies have shown that code-switching data (CSD), in which multiple languages are mixed within the same context, can improve cross-lingual transfer and multilingual alignment in large language models (LLMs). However, existing studies primarily focus on bilingual transfer between English and a target language, leaving multilingual settings involving three or more languages largely unexplored. In this work, we investigate multilingual code-switching instruction tuning across four languages: English, Japanese, Korean, and Chinese. We evaluate multilingual understanding on Belebele. Our experiments show that simple sentence-level multilingual CSD consistently improves average multilingual performance across all four languages, indicating that multilingual code-switching can be effective beyond bilingual transfer settings.

160. Semantic and Visual Evidence for Efficient Long-Video Reasoning: A Solution for the HD-EPIC VQA Challenge

Authors: Yinsong Xu , Wei Jing , Liuxin Zhang , Wanjun Lv , Hui Li
URL: https://arxiv.org/abs/2605.29402
Abstract:

Understanding long-form egocentric videos remains challenging for multimodal large language models (MLLMs) due to limited context length and insufficient grounding of fine-grained visual details. The recently proposed HD-EPIC benchmark highlights these limitations: even strong long-context models achieve relatively low performance across diverse video question answering tasks. In this paper, we propose a unified framework that decouples long-video reasoning into two complementary forms of evidence: semantic evidence and visual evidence. Semantic evidence captures global procedural structure through a coarse-to-fine extraction pipeline, while object-centric visual evidence preserves fine-grained grounding through bounding boxes and visual embeddings. During inference, we formulate reasoning as a query-conditioned evidence retrieval and integration process, dynamically selecting relevant information from both sources. Our approach achieves competitive performance in the HD-EPIC-VQA Challenge across multiple task categories. More broadly, our results demonstrate that explicitly structuring, retrieving, and integrating semantic and visual evidence is critical for effective long-video understanding with MLLMs.

161. GDSD: Reinforcement Learning as Guided Denoiser Self-Distillation for Diffusion Language Models

Authors: Xiaohang Tang , Keyue Jiang , Che Liu , Qifang Zhao , Xiaoxiao Xu , Sangwoong Yoon , Ilija Bogunovic
URL: https://arxiv.org/abs/2605.29398
Abstract:

Reinforcement learning (RL) can be used to improve the policy (denoiser) of diffusion large language models (dLLMs), while being hindered by the intractability of the policy likelihood. A dominant and efficient family of methods replaces the likelihood in standard RL with its evidence lower bound (ELBO), estimated from randomly masked sequences. Despite being well aligned with pre-training, these approaches introduce bias through training–inference mismatch by using the ELBO as a likelihood surrogate, which can degrade performance. In this work, we propose Guided Denoiser Self-Distillation (GDSD) to directly distill the denoiser of dLLMs from an advantage-guided self-teacher, derived from the closed-form optimum of reverse-KL regularized RL. GDSD matches the dLLM’s denoiser logits to the teacher’s via a normalization-free objective, which reduces RL to likelihood-free self-distillation and thus bypasses the TIM biases. Recent ELBO-based methods emerge as instances of applying different distillation divergences, but with diagnosable pathologies that GDSD avoids. On planning, math, and coding benchmarks with LLaDA-8B and Dream-7B, GDSD consistently outperforms prior state-of-the-art ELBO-based methods with a more stable training reward dynamics, achieving test-accuracy improvements of up to $+19.6\%$. These results suggest that direct denoiser self-distillation, without relying on an ELBO likelihood surrogate, can provide a more stable and effective RL procedure for dLLMs. Code is available at this https URL .

162. On the Optimizer Dependence of Neural Scaling Laws

Authors: Vansh Ramani , Shourya Vir Jain
URL: https://arxiv.org/abs/2605.29387
Abstract:

The scaling exponent $\alpha$ in neural scaling laws $L(N) \propto N^{-\alpha}$ is commonly treated as a fixed constant set by architecture and data. We present evidence that $\alpha$ depends systematically on the optimizer. In controlled random-feature regression experiments – the canonical theoretical framework for neural scaling – we measure $\alpha$ across five optimizer variants and six spectral conditions. Preconditioned optimizers consistently yield steeper scaling (larger $\alpha$), with the $\alpha$-shift increasing across most of the tested spectral range, peaking near $s = 1.5$, and remaining large at $s = 2.0$. At $s \approx 1.0$ (characteristic of natural language), the full natural gradient achieves $\alpha \approx 0.31$ versus $\alpha \approx 0.12$ for gradient descent – a $2.6\times$ larger fitted exponent that, within the random-feature model, compounds with each model-size doubling. Whether and how this exponent shift transfers to large-scale LLM training – where recent evidence suggests the advantage may attenuate with scale – remains an important open question. Our results imply that scaling-law forecasts should account for optimizer choice, and we provide a spectral diagnostic predicting when advanced optimizers will pay off.

163. SURGENT: A Surgical Multi-Agent Assistance System Across the Perioperative Workflow

Authors: Dongsheng Shi , Yue Li , Xin Yi , Yongyi Cui , Huawei Feng , Linlin Wang
URL: https://arxiv.org/abs/2605.29368
Abstract:

The intricate nature of modern surgical care necessitates intelligent systems that can synthesize extensive patient records, support collaborative decision-making, and provide transparent, auditable reasoning across the entire perioperative workflow. Although web-based Large Language Models (LLMs) possess advanced reasoning capabilities, they are ill-equipped for surgical applications due to critical limitations: input length constraints, incomplete memory management, and limited traceability. To address this issue, we present SURGENT, a surgical multi-agent assistance system that combines a Tree-of-Thought planner, multi-department collaboration agents, and retrieval-augmented reasoning with clinical guidelines and biomedical literature. SURGENT features a novel memory design that manages both long-term patient histories and short-term working summaries, enabling more complete, contextualized, and consistent reasoning. Experimental evaluations across five key perioperative tasks - case analysis, surgical plan simulation, safety monitoring, complication risk assessment, and rehabilitation guidance - show that SURGENT outperforms baseline LLMs and existing medical multi-agent frameworks, yielding recommendations more closely aligned with patient histories. Ablation studies further highlight the advantage of DeepSeek as a locally deployable backbone model, enabling privacy-preserving deployment without reliance on centralized services. These results position SURGENT as a practical and trustworthy advancement toward intelligent, equitable, and secure surgical assistance systems.

164. GrepSeek: Training Search Agents for Direct Corpus Interaction

Authors: Alireza Salemi , Chang Zeng , Atharva Nijasure , Jui-Hui Chung , Razieh Rahimi , Fernando Diaz , Hamed Zamani
URL: https://arxiv.org/abs/2605.29307
Abstract:

Large Language Model (LLM) search agents have shown strong promise for knowledge-intensive language tasks through multiple rounds of reasoning and information retrieval. Most existing systems access information using a retriever that takes a keyword or natural language query and returns a ranked list of documents using an index of pre-computed document representations. In this work, we explore a complementary perspective in which the search agent treats the corpus itself as the search environment and finds evidence by issuing executable shell commands. We introduce GrepSeek, an optimized direct corpus interaction (DCI) search agent that trains a compact search agent to find, filter, and compose evidence from large text corpora. To address the instability of learning behavior directly with reinforcement learning on large corpora, we propose a two-stage training pipeline. First, we construct a cold-start dataset using an answer-aware Tutor and answer-blind Planner to generate verified, causally grounded search trajectories. Second, we refine the initialized policy with Group Relative Policy Optimization (GRPO), allowing the agent to improve its task-oriented search behavior through direct interaction with the corpus. To make DCI practical at scale, we further use a semantics-preserving sharded-parallel execution engine that accelerates shell-based retrieval by up to $7.6\times$ while preserving byte-exact equivalence with sequential execution of the shell command. Experiments across seven open-domain question answering benchmarks show that GrepSeek achieves the strongest overall token-level $F_1$ and Exact Match. Our analysis also highlights the limitations of purely lexical interaction on queries with substantial surface-form variation, suggesting DCI as a practical and competitive method for search agents that can complement existing retrieval paradigms in the real world.

165. MusTBENCH: Benchmarking and Advancing Temporal Grounding in Music LLMs

Authors: Daeyong Kwon , Qiyu Wu , Shinobu Kuriya , Junghyun Koo , Shuyang Cui , Zhi Zhong , Wei-Hsiang Liao , Hiromi Wakaki , Yuki Mitsufuji
URL: https://arxiv.org/abs/2605.29300
Abstract:

Recent Large Audio-Language Models (LALMs) have demonstrated promising abilities in understanding musical content. However, whether their responses are grounded in the correct temporal regions of the audio remains underexplored. This limitation is particularly critical for music understanding, where key information often occurs as temporally localized events, such as instrument entries and rhythmic transitions. To address this gap, we introduce MusTBENCH, a music-expert-validated benchmark designed to evaluate temporal grounding in LALMs through five temporally grounded question-answering tasks. To further improve temporal grounding in existing models, we propose MusT, a novel four-stage temporal optimization recipe spanning music encoder adaptation, LLM adaptation, LLM supervised fine-tuning, and RL-based optimization. Experiments on MusTBENCH show that existing LALMs struggle with precise temporal grounding, while MusT brings significant improvements over strong baselines. These results establish temporal grounding as a key missing capability in current LALMs and position MusTBENCH as a challenging benchmark for future research in temporally grounded music understanding.

166. Pocket-Dentist: On-Device Dental Image Understanding via Efficient Multimodal Large Language Models

Authors: Kai Bian , Xucheng Guo , Bin Chen , Lingyan Ruan , Yiran Shen , Ting Dang , Hong Jia
URL: https://arxiv.org/abs/2605.29299
Abstract:

Evaluations of dental vision-language models remain fragmented across datasets, task definitions and metrics, and often ignore their computational cost. This limits their widespread deployment for dental screening outside specialist centres, where timely inference, limited hardware, and local handling of patient images are vital for practical, privacy-preserving clinical prescreening. Here we present Pocket-Dentist, an efficiency-aware benchmark for dental multimodal question answering that brings together three datasets spanning approximately 1,159 patients, five task types and seven metrics. Across typical 14 VLMs, our results reveals an interesting observation: compact VLMs (e.g., 2B-parameter models) outperform larger VLMs in accuracy while requiring substantially lower computational costs in dental image understanding. Deployed locally on an iPhone 17 Pro, our finetuned compact VLM Pocket-Dentist-2B processed each sample in 4.31 s, reducing latency by 4.9-fold and memory use by 2.3-fold compared with a 7B baseline.

167. Code-QA-Bench: Separating Code Reasoning from Documentation Memorization in Repository-Level QA

Authors: Jun Zhang , JianYing Qu , Hanwen Du , Zhongkai Sun , Yehua Yang , Qiao Zhao
URL: https://arxiv.org/abs/2605.29277
Abstract:

We present Code-QA-Bench, a fully automated framework for synthesizing repository-level code understanding benchmarks that separates genuine code comprehension from documentation recall and pretraining memorization. The framework makes two methodological contributions: (1) an answer-first generation pipeline where a tool-equipped agent explores source code to produce verified gold answers before deriving questions, ensuring every task is grounded in real code structure; and (2) a three-condition experimental design evaluating agents under closed-book (no repository), code-only (documentation removed), and documented (full repository) conditions, with deltas directly quantifying documentation utility and memorization. We generate 528 code-derivable and 100 doc-dependent tasks across 10 Python repositories from SWE-Bench, scored by an LLM judge on accuracy, completeness, and specificity. Experiments on four frontier models reveal that code access is the dominant factor (+0.23 mean gain over closed-book), documentation provides modest additional benefit (+0.071 on doc-dependent tasks), and code-only $\approx$ documented on code-derivable tasks, validating the design. The framework is open-source and applicable to any well-documented Python repository.

168. Compute Allocation in Evolutionary Search: From Depth-Breadth to Multi-Armed Bandits

Authors: Sixue Xing , Haoyu He , Kerui Wu , Zhuo Yang , Haozheng Luo , Tianfan Fu , Aarthy Nagarajan
URL: https://arxiv.org/abs/2605.29268
Abstract:

LLM-guided evolutionary search (Evolve systems) has reached state-of-the-art results on mathematical and combinatorial tasks, yet most existing systems report only the best of many runs and leave the run-to-run distribution undocumented. We ask how a fixed budget of LLM calls should be allocated, and how reliably a single run reaches the reported numbers. Sweeping the depth-breadth grid over five models and three tasks, we identify two empirical regularities: a fitness-compute envelope along which capability ordering largely collapses on effective FLOPs, and a bilinear depth-breadth fit with task-specific interaction; both are gated by model-task capability. Motivated by these regularities, we propose BaSE (Bandit-based Self-Evolving), a multi-armed bandit that allocates LLM calls across parallel trajectories. Without changing the model, prompt, or evaluator, BaSE improves mean fitness by 12.3% over the strongest island-protocol baseline across 8 (model, task) cells, with the largest gains on high-variance settings: a reliability gain from allocation alone.

169. DynSess: Dynamic Session-Level Evaluation and Optimization Framework for Role-Playing Agents

Authors: Rongsheng Zhang , Jiji Tang , Junnan Ren , Zuyi Bao , Weijie Chen , Ruofan Hu , Zhou Zhao , Tangjie Lv , Yan Zhang
URL: https://arxiv.org/abs/2605.29256
Abstract:

Role-playing with large language models is fundamentally a session-level task, requiring agents to sustain character identity and interaction quality across extended multi-turn conversations. Yet existing evaluation and optimization methods remain largely turn-level, failing to capture long-horizon quality. We propose DynSess, a unified session-level framework for role-playing agents. DynSess-Eval scores complete dialogue sessions via rubrics targeting long-horizon behaviors. Leveraging its session-level rewards, we construct high-quality training trajectories through multi-turn lookahead search and train DynSess-Character with two complementary variants: DSPO (off-policy) and GSRPO (on-policy). Experiments show that DynSess-Eval aligns with human judgments substantially better than prior evaluators, and blind human evaluation further shows that DynSess-Character matches the strongest character model despite using substantially fewer parameters, while maintaining strong role consistency and interactive ability. Our dataset and code will be released to facilitate future research.

170. BlockBatch: Multi-Scale Consensus Decoding for Efficient Diffusion Language Model Inference

Authors: Xiaoyou Wu , Cheng-Jhih Shih , Binfei Ji , Yong Liu , Yingyan (Celine)Lin
URL: https://arxiv.org/abs/2605.29233
Abstract:

Diffusion language models (dLLMs) generate text by iteratively denoising multiple token positions in parallel, offering an attractive alternative to strictly autoregressive decoding. In practice, however, block-wise dLLM inference exposes a difficult granularity trade-off: small blocks preserve local conditioning but require many denoising steps, whereas large blocks expose more parallelism but can make premature commitments and accumulate cache error. Existing acceleration methods typically choose a single block size per request, leaving the complementarity among block sizes unused. We show that block size itself is a useful branching dimension. Different block sizes induce related but non-identical KV-cache trajectories: branches often share an initial prefix, bifurcate at semantically decisive positions, and later agree on syntactically lightweight tokens. Motivated by this structure, we propose BlockBatch, a training-free online inference framework that executes multiple block-size branches for the same request inside a batched forward pass. BlockBatch coordinates these branches through confidence-gated token merging, leader-based synchronization, and periodic full-sequence refreshes that re-anchor local block updates to a globally consistent KV state. Across 3 representative dLLMs and 4 datasets, BlockBatch reduces denoising NFEs by 26.6\% on average and achieves a 1.33$\times$ average end-to-end speedup over Fast-dLLM while preserving accuracy. These results identify block-size diversity as a practical and previously underexplored axis for branch-parallel dLLM inference.

171. Relevance as a Vulnerability: How Web Retrieval Degrades Safety Alignment in LLM Agents

Authors: Aditya Nawal , Manit Baser , Mohan Gurusamy
URL: https://arxiv.org/abs/2605.29224
Abstract:

AI agents augment large language models with external tools such as web retrieval, enabling grounded and up-to-date responses. However, incorporating external content into the generation pipeline can weaken the safety alignment mechanisms that govern model outputs. Prior work shows that enabling retrieval in agents increases compliance with harmful requests. We introduce AgentREVEAL, a diagnostic framework for analyzing retrieval-induced safety degradation in LLM agents. The framework examines two axes: how retrieval is integrated into the agent pipeline and the properties of the retrieved content. Along the integration axis, we find that binding tool invocation and response generation in a single step amplifies harmful outputs. Along the content axis, we uncover the Safe Source Paradox: even oppositional or safety-oriented sources, such as pages containing warnings or risk disclaimers, can increase harmful compliance by an average of 25% compared to the no-retrieval baseline. Finally, we show that relevance acts as a shared activation condition for both vulnerabilities. Similar patterns appear on frontier closed models, and harmful compliance remains elevated under several representative pipeline interventions, with some agents also entering this regime under autonomous retrieval. Because relevance is also what makes retrieval useful, these results expose a safety-utility trade-off for retrieval-enabled agents. We introduce HarmURLBench, a benchmark containing 1,405 real-world URLs paired with 320 harmful behaviors to support future evaluations.

172. Influence-Guided Symbolic Regression: Scientific Discovery via LLM-Driven Equation Search with Granular Feedback

Authors: Evgeny S. Saveliev , Samuel Holt , Nabeel Seedat , David L. Bentley , Jim Weatherall , Mihaela van der Schaar
URL: https://arxiv.org/abs/2605.29184
Abstract:

Large Language Models (LLMs) offer a promising avenue for scientific discovery, yet their application to symbolic regression is often constrained by inefficient search strategies and coarse feedback signals. Current methods typically guide LLMs using scalar metrics (e.g., global Mean Squared Error), which fail to identify which components of a proposed equation are driving performance or causing error. We introduce \textit{Influence-Guided Symbolic Regression} (IGSR), a method that frames equation discovery as an iterative two-step process combining diverse term generation with rigorous selection: an LLM generates candidate basis functions $\psi_j(\mathbf{x})$ for a linear model, which are then evaluated using granular influence scores $\Delta_j$. These scores quantify each term’s marginal contribution to generalization accuracy, enabling an influence-guided pruning process that systematically refines the model structure. Integrating this mechanism into a Monte Carlo Tree Search (MCTS) enables navigating the combinatorial search space while balancing exploration of novel functional forms with exploitation of high-influence components. We demonstrate IGSR’s effectiveness on a diverse suite of benchmarks, including LLM-SRBench, pharmacological PKPD models, an epidemiological simulation, and real-world genomic data. Notably, we validate the framework’s capacity for genuine discovery in a case study using a high-dimensional biological dataset, in which IGSR identified a novel relationship between DNA methylation and RNA Polymerase II pausing; a hypothesis that was subsequently supported via wet-lab experimentation.

173. Sustainable Metal-Organic Framework Water Harvesters in the Artificial Intelligence Era

Authors: Reid A. Coyle (1), Shyam Chand Pal (1), Peter Walther (1), Saeun Park (1), Bin Feng (1,2), Zhiling Zheng (1,2) ((1) Department of Chemistry, Washington University, St. Louis, MO, United States, (2) Institute of Materials Science & Engineering, Washington University, St. Louis, MO, United States)
URL: https://arxiv.org/abs/2605.29179
Abstract:

Metal-organic frameworks (MOFs) are excellent candidates for water harvesting due to their tunable pore environments, which can be precisely engineered to capture and release water in arid conditions. Integrating artificial intelligence (AI) into MOF discovery can further accelerate the design of high-performance sorbents by identifying structural features that enhance atmospheric water harvesting (AWH), stability, and cycling efficiency. In this Perspective, we examine key MOF design principles, including cooperative adsorption, operational relative humidity (RH), uptake capacity, hysteresis, and scalability. We highlight recent design advancements such as multivariate strategies and long-arm linker extension, and examine how these principles tune pore capacity and hydrophilicity, while preserving stability and crystallinity. Furthermore, we discuss how AI, large language models (LLMs), and data mining can accelerate the discovery process through predictive synthesis, inverse design, and elucidating synthesis-structure-property relationships for the next generation of MOF water harvesters.

174. UA-Legal-Bench: A Benchmark for Evaluating Large Language Models on Ukrainian Legal Reasoning

Authors: Volodymyr Ovcharov
URL: https://arxiv.org/abs/2605.29170
Abstract:

Legal NLP benchmarks are overwhelmingly English-centric, leaving failure modes in morphologically rich, non-Latin-script languages undetected. We introduce UA-Legal-Bench, a five-task benchmark for evaluating large language models on Ukrainian legal reasoning, built from the Unified State Register of Court Decisions (EDRSR) – one of the world’s largest open judicial corpora (99.5 million decisions). The benchmark comprises: (1) case-type classification (4 classes, n=2,000), (2) judgment form classification (4 classes, n=2,000), (3) case-outcome prediction (6 classes, n=800), (4) legal norm extraction (n=1,794), and (5) cause category prediction (22 classes, n=1,871). We evaluate 11 LLMs (3B–675B) from five families under zero-shot and 3-shot prompting via AWS Bedrock with 158K API calls. Our results reveal sharply task-dependent few-shot effects: few-shot prompting improves judgment form classification by up to +38.6 pp but has mixed effects on outcome prediction. We show that accuracy is misleading on imbalanced legal tasks: the model with highest COP accuracy (62%) is a majority-class predictor (macro-F1: 23%), while the genuinely best model scores only 44% macro-F1. Within-family scaling analysis reveals that 8B models can match frontier performance on surface-level tasks but scaling thresholds vary dramatically across families. We release all data, prompts, and model predictions.

175. Parallax: Parameterized Local Linear Attention for Language Modeling

Authors: Yifei Zuo , Dhruv Pai , Zhichen Zeng , Alec Dewulf , Shuming Hu , Zhaoran Wang
URL: https://arxiv.org/abs/2605.29157
Abstract:

Large Language Models (LLMs) have become the central paradigm in artificial intelligence, yet the core computational primitive of attention has remained structurally unchanged. Local Linear Attention (LLA) is an attention mechanism derived from nonparametric statistics in the test-time regression framework. In contrast to prior research on efficient attention variants, LLA upgrades the local constant estimate in softmax attention to a local linear estimate, yielding provably superior bias-variance tradeoffs for associative memory. However, LLA has not been scaled in LLM pretraining due to computational and numerical stability concerns. We introduce Parallax, a parameterized Local Linear Attention that is scalable for LLMs. Parallax eliminates the numerical solver in LLA and learns an extra query-like projector that probes the KV covariance. We place Parallax within a family of attention mechanisms connected by the bandwidth, the probe construction and the affine structure. We propose a hardware-aware algorithm that increases the arithmetic intensity over FlashAttention, shifting attention into a more compute bound regime. Our prototype decode kernel matches or outperforms FlashAttention 2/3 across diverse batch sizes and context lengths. We pretrain Parallax at 0.6B and 1.7B scales and find consistent perplexity improvements throughout pretraining with gains that transfer to downstream benchmarks. The advantage persists under both parameter-matched and compute-matched controls, demonstrating a Pareto improvement. We perform careful pretraining ablations and identify a novel phenomenon whereby Muon unlocks the capacity of Parallax. To our knowledge, this is the first empirical demonstration of strong architecture-optimizer codesign for attention mechanisms in the architecture research literature.

176. SafeRx-Agent: A Knowledge-Grounded Multi-Agent Framework for Safe and Explainable Medication Recommendation

Authors: Xinyu Wang , Hanwei Wu , Zhenghan Tai , Sicheng Lyu , Qincheng Lu , Ziyu Zhao , Jijun Chi , Jingrui Tian , Xiao-Wen Chang , Ziyang Song
URL: https://arxiv.org/abs/2605.29146
Abstract:

Medication recommendation predicts medications for patient visits, but existing methods still face two key challenges. At the model level, traditional drug recommendation methods only predict structured drug codes with limited evidence grounding, while LLM agents can use richer clinical context but may lack safety verification and traceability. At the task level, existing benchmarks often use broad medication categories, which ignore subgroup-level safety differences and can lead to risk overestimation. We introduce the first fine-grained medication recommendation setting based on fourth-level ATC code generation. We propose Safe Prescription Agent (SafeRx-Agent), a knowledge-grounded multi-agent framework that uses patient context, external clinical knowledge, and safety verification to recommend traceable medication sets. Experimental results on MIMIC-III and MIMIC-IV datasets show that SafeRx-Agent improves fine-grained medication prediction accuracy while controlling drug interactions, contraindications, and medication set size.

177. Toward User Preference Alignment in LLM Recommendation via Explicit Context Feedback

Authors: Weizhi Zhang , Wooseong Yang , Yuxin Cui , Zhaohui Guo , Hins Hu , Liangwei Yang , Henry Peng Zou , Qifei Wang , Hanqing Zeng , Jiayi Liu , Yinglong Xia , Philip S. Yu
URL: https://arxiv.org/abs/2605.29141
Abstract:

Traditional recommender systems (RecSys) primarily infer user preferences from implicit signals (such as clicks, watches, and purchases), often neglecting the rich explicit contextual feedback users provide through verbal text, like comments and reviews. This explicit context feedback captures the nuanced reasons behind user decisions regarding their preferences. In addition, it offers critical heterogeneous information for user preference alignment and more explainable recommendations. Overlooking such signals can lead to misaligned user preferences and further reinforce filter bubbles, as algorithms fail to understand the “semantic context” behind user choices. Recent advances in Large Language Models (LLMs) present new opportunities to harness user-generated content for more accurate and diverse recommendations, yet current LLM-based recommendations still focus on using item meta-data and underutilize this resource. In this paper, we advocate for prioritizing explicit context feedback in the next generation of LLM-based RecSys. We review the evolution of recommendation paradigms, highlight the value of context-rich feedback, call for new benchmarks and metrics, and introduce frameworks for integrating explicit user signals into scalable LLM-driven RecSys. Centering on user-preference modeling, we aim to foster more personalized, transparent, and explainable RecSys online platforms.

178. When and How Long? The Readout-Mediator Angle in Temporal Reasoning

Authors: Shreyas Fadnavis , Praitayini Kanakaraj , Felix Wyss
URL: https://arxiv.org/abs/2605.29126
Abstract:

A linear probe can decode a representation almost perfectly and yet be completely irrelevant to how the model uses it. On calendar-date duration reasoning in language models, a $\sin$/$\cos$ probe recovers day-of-year from a layer’s activations, yet ablating its direction has no effect on the model’s answers – while ablating a four-dimensional subspace found by Distributed Alignment Search (DAS) at the same layer collapses performance entirely. We measure the angle between these two subspaces – the \emph{readout-mediator angle} – and find it indistinguishable from the angle between two random subspaces (the Haar-uniform null), meaning the probe has learned a direction orthogonal to the model’s actual computation. Reverse-engineering the circuit reveals why: attention heads route month-grained context through learned QK offsets at ${\pm}30$ and ${\pm}61$ days, and MLPs then convert \emph{when} (absolute date) into \emph{how long} (duration) – all downstream of the causal subspace the probe never touches. Sparse-autoencoder decomposition confirms the split: probe-aligned and DAS-aligned features encode semantically disjoint concepts with negligible causal overlap. The dissociation replicates across four scales ($1.5$-$9\,$B) and two model families, with preliminary evidence on two further domains (spatial displacement, symbolic arithmetic), suggesting that readout-mediator orthogonality is a general failure mode of probe-based interpretability. This directly undermines proposals to deploy probes as runtime safety monitors: the probe can report high confidence on a direction the model has silently abandoned.

179. unix-ctf: Procedural Environments for Unix-Competence Reinforcement Learning

Authors: Geoffrey Bradway , Roger Creus Castanyer , Lorenz Wolf , Maxwill Lin , Matthew James Sargent , Augustine N. Mavor-Parker
URL: https://arxiv.org/abs/2605.29115
Abstract:

Unix competence is the ability to use shell and operating-system primitives as first-class tools, not merely to write programs through a terminal. Current terminal benchmarks tend to blur this distinction: a solver fluent in Python but weak in Unix can pass a substantial fraction of Terminal-Bench 2.0, while the reverse skill profile is rarely exercised. We make the distinction operational and build a training surface for the Unix component. unix-ctf is a procedural generator of capture-the-flag tasks for shell agents. Each task hides a short token (a flag of the form flag(a3b1c9…)) inside a fresh Linux container using a single Unix feature, and the agent must recover it. Tasks are produced by an LLM-assisted synthesis pipeline that generates candidate hiding techniques, rewrites them into parameterized hide-and-find script pairs, and filters them with a bidirectional contract: the hide script must leave no plaintext trace of the flag on disk, and the find script must recover the flag in a fresh directory. Because the LLM only writes the planting and recovery steps (the container, layout, and grading harness are fixed), the pipeline lands 656 of 750 raw attempts as portable, reusable variants (87.5\%). Our reproduction of Endless Terminals’ full-container-generation approach lands only 17.4\% under the same checks. The 656 variants canonicalize to 155 distinct techniques. Fine-tuning Qwen3-8B with LoRA using GRPO on this surface lifts solve rate from 11.6\% to 43.6\% on a 15-skill multi-family holdout (n=225), redistributes which InterCode-CTF tasks the model solves, and produces a +33 pp gain in Forensics while reaching 32/100 on InterCode-CTF. These results suggest that Unix competence is separable, trainable, and best evaluated directly rather than folded into programming-through-a-shell.

180. GEO-Bench: Benchmarking Ranking Manipulation in Generative Engine Optimization

Authors: Ojas Nimase , Zhe Chen , Gengpei Qi , Yue Zhao , Xiyang Hu
URL: https://arxiv.org/abs/2605.29107
Abstract:

Large language models (LLMs) increasingly rank products, documents, and recommendations for user queries, which makes manipulating these rankings a growing concern for fairness and information integrity. Research on generative engine optimization (GEO) has produced many manipulation methods, but each is evaluated on its own dataset with its own metrics, so their relative strength and detectability stay unclear. We present GEO-Bench, a benchmark that evaluates GEO ranking-manipulation attacks under one protocol. It unifies black-box prompt-based attacks (TAP, Zero-Shot), white-box gradient-based attacks (STS, RAF, StealthRank), and ten white-hat C-SEO strategies. We score every method on five datasets against a fixed open-weight ranker (Llama-3.1-8B-Instruct), using metrics for both effectiveness (NRG, Success@{\alpha}, Promote@{\alpha}) and stealth (keyword violation rate, perplexity ratio). Our evaluation shows that effectiveness and stealth trade off across adversarial attacks, that black-box content rewriting matches or exceeds gradient-based attacks on rank promotion while producing more fluent text and can evade both keyword- and perplexity-based detection on some domains, and that the access model does not predict attack strength. By standardizing datasets, attack implementations, and metrics, GEO-Bench enables the first direct comparison across these attack paradigms and supports the development of detection methods.

181. OISD: On-Policy Internal Self-Distillation of Language Models

Authors: Xinyu Liu , Darryl Cherian Jacob , Yang Zhou , Jindong Wang , Pan He
URL: https://arxiv.org/abs/2605.29089
Abstract:

Recent reinforcement learning (RL) post-training approaches primarily optimize the final output policy using sparse outcome-level rewards, while largely overlooking predictive signals encoded in intermediate representations. In this paper, we introduce a new paradigm called on-policy internal self-distillation and propose the OISD framework, which improves reasoning by transferring on-policy predictive signals from the final layer to intermediate representations. During rollout and Group Relative Policy Optimization (GRPO) optimization, the final layer acts as both the policy and a detached internal teacher for selected intermediate layers, which are guided to align with it through two complementary mechanisms: logit alignment, which transfers high-level reasoning behaviors (how to think), and attention alignment, which enforces consistent attention patterns (where to look) from the final layer to the selected intermediate layer, both without requiring external privileged information. Our OISD, together with GRPO, employs signed advantage-weighted Jensen–Shannon alignment to distill informative intermediate representations while preserving policy consistency under a unified acting policy. Experimental results demonstrate the effectiveness of OISD, with substantial and consistent improvements over strong reasoning RL baselines across four mathematical reasoning tasks. The code will be released at this https URL

182. Structured Prompt Optimization Meets Reinforcement Learning for Global and Local Interpretability over Complex Text

Authors: Tianyang Zhou , Wenbo Chen , Pierre Jinghong Liang , Leman Akoglu
URL: https://arxiv.org/abs/2605.29076
Abstract:

LLMs have advanced text classification, yet existing paradigms face a trade-off: supervised (label only) fine-tuning is scalable but offers limited reasoning on complex text and lacks broader model transparency, while discrete prompt optimization offers human-readable instructions but struggles with performance and scalability. We introduce eXTC (eXplainable Text Classifier) with three progressive stages: (1) learning a Standard Operating Procedure (SOP, or rulebook) in natural language via a new Structured Prompt Optimization algorithm; (2) SOP-grounded reasoning distillation from a large teacher LLM into a compact LM; and (3) expanding reasoning capabilities beyond the initial SOP via reinforcement learning. This design enables eXTC to provide (i) fast inference via a compact LM, with (ii) inference-time local reasoning traces, alongside a global, modular explanation of its learned domain rules, while (iii) significantly outperforming existing paradigms across diverse benchmarks in both classification performance and explanation quality, with stage-by-stage gains.

183. SCDBench: A Benchmark for LLM-Based Smart Contract Decompilers

Authors: Kaihua Qin , Dawn Song , Arthur Gervais
URL: https://arxiv.org/abs/2605.29059
Abstract:

Smart contract decompilation aims to recover high-level source code from bytecode, but evaluating decompilers remains difficult because existing studies use narrow datasets, inconsistent metrics, and limited semantic consistency checks. This gap is increasingly important as large language models (LLMs) begin to generate source-like Solidity that may compile and appear plausible, even when its semantics diverge from the original contract. We introduce SCDBench, a dataset and benchmark methodology for LLM-based smart contract decompilation. The dataset contains 600 real-world Solidity contracts with paired bytecode inputs, ground-truth source code, and replayable semantic checkpoints. SCDBench evaluates decompiler outputs through four cumulative stages: format completeness, compilability, Application Binary Interface (ABI) recovery, and semantic consistency via differential replay. We evaluate Claude Opus 4.7, GPT-5.3-Codex, and GLM-5 in a zero-shot decompilation setting, including GLM-5 variants with and without extended reasoning and a zero-shot compilation-repair setting. The results show that frontier LLMs can often produce structured and compilable Solidity, but achieving semantic consistency remains far from solved: the best-performing frontier model perfectly decompiles only 42/600 contracts. We further show that introducing same-model compilation repair substantially improves performance at modest additional cost. SCDBench establishes a common ground for rigorous, reproducible evaluation and aims to accelerate the development of reliable smart contract decompilers for blockchain security and transparency.

184. Label-Free Reinforcement Learning via Cross-Model Entropy

Authors: Matt Gorbett , Hossein Shirazi
URL: https://arxiv.org/abs/2605.29009
Abstract:

Post-training large language models with reinforcement learning is bottlenecked by the reward signal. Existing approaches require either ground-truth verifiable rewards, restricting training to domains with automatic correctness checks (e.g., mathematics, code execution), or human preference labels, which are expensive to collect and prone to reward hacking. Recent label-free methods replace ground-truth verifiers with self-referential signals like majority voting or token entropy over a model’s own outputs, but risk reinforcing a model’s own errors. In this work we propose Cross-Model Entropy (CME), the mean log-likelihood of a generator’s response under a separate verifier model, as a label-free reward signal for RL post-training. CME is continuous, training-free, and grounded in the principle that responses a verifier finds unsurprising are likely correct or high quality. Because the verifier is independent of the generator, the signal cannot be gamed through self-consistency. We integrate CME into GRPO with no other changes to the training loop, extending label-free RL to open-ended instruction following – a regime where self-referential signals are inapplicable or poorly suited. On open-ended instruction following (UltraFeedback prompts, evaluated on AlpacaEval 2.0), CME rewards beat the untrained base in head-to-head LLM-as-Judge comparisons across four model families (Qwen, Llama, Gemma, OLMo) and three training regimes (pretrained, SFT, and instruction-tuned), with tie-adjusted win rates ranging from 52.5% to 71.4%. Code will be released upon publication.

185. Measuring Real-World Prompt Injection Attacks in LLM-based Resume Screening

Authors: Mohan Zhang , Yuqi Jia , Zhen Tan , Steven Jiang , Neil Zhenqiang Gong , Tianlong Chen , Dawn Song
URL: https://arxiv.org/abs/2605.28999
Abstract:

LLMs are vulnerable to prompt injection attacks. However, this vulnerability has been primarily demonstrated conceptually in academic studies or through a few anecdotal case studies. Its prevalence and impact in real-world LLM-based applications are largely unexplored. In this work, we present the first systematic study of prompt-injection attacks in a widely used application: LLM-based resume screening. Our analysis is based on approximately 200K real-world resumes collected over multiple years by hireEZ. We first design tailored methods to detect prompt injection in resumes. Manual validation on a small-scale dataset demonstrates that our detectors achieve high precision and outperform state-of-the-art general-purpose detectors. We then apply our detector to the full resume dataset and conduct a comprehensive measurement study of real-world prompt injection attacks. Our analysis reveals several intriguing findings: approximately 1% of resumes contain hidden prompt injections; the prevalence of such injected resumes has increased noticeably over the past one to two years; and more than 90% of injected prompts do not use explicit instructions. These results provide the first evidence of large-scale prompt injection in real-world LLM-based applications and lay the groundwork for future studies to understand and mitigate such attacks.

186. Beyond Recall: Behavioral Specification as an Interpretive Layer for AI Personalization

Authors: Aarik Gulaya
URL: https://arxiv.org/abs/2605.28969
Abstract:

If an AI agent makes decisions on a person’s behalf, those decisions must align with its user. We introduce representational accuracy to measure how faithfully a system captures a person’s interpretation. An interpretive layer is operationalized as a Behavioral Specification. Our reference implementation aggressively compresses a person’s data into interpretive patterns, served as context to a language model. We evaluate the Specification on a prototype benchmark of held-out behavioral predictions scored by a calibrated 5-judge LLM panel. We test it independently and in composition with a range of context conditions: full raw corpus, full extracted facts, and four commercial memory systems (Mem0, Letta, Supermemory, Zep). Across 14 public-domain autobiographical corpora, the Specification lifts representational accuracy in aggregate and nearly eliminates model hedging. It recovers most of what the raw corpus delivers, at ~25x less context cost. The Specification lifts subjects toward a common predictive level regardless of pretraining baseline; the lift in absolute points is therefore largest where the baseline is lowest, suggesting the population of relevance is anyone not adequately represented in pretraining. Lift is greatest on interpretation-required questions, where providing an interpretive layer enables model behavior that extracted facts or raw corpus do not. Conversely, on recall-required questions, this layer can interfere rather than help. We conclude that representational accuracy is distinct from recall and that human-AI alignment is dependent on how accurately the user is represented. Representational accuracy makes that alignment testable.

187. Conf-Gen: Conformal Uncertainty Quantification for Generative Models

Authors: Gabriel Loaiza-Ganem , Kevin Zhang , Wei Cui , Marc T. Law , Kin Kwan Leung
URL: https://arxiv.org/abs/2605.28920
Abstract:

Conformal prediction (CP) and its extension, conformal risk control (CRC), are established frameworks for quantifying uncertainty in supervised machine learning through formal guarantees. However, recent breakthroughs in artificial intelligence (AI) have been driven by unsupervised generative models, such as large language models (LLMs) and image generators, which are not directly compatible with CP or CRC. In this work we introduce conformal generation (Conf-Gen), a general framework adapting CRC to generative tasks while relaxing its theoretical assumptions. Conf-Gen unifies and generalizes previous attempts to apply CP to LLMs, and extends conformal methodology to entirely new domains. We demonstrate the flexibility of Conf-Gen through some novel applications, including obtaining conformal guarantees on: image generators producing non-memorized images, conversational AI systems having asked enough clarifying questions, and the output of AI agents being correct.

188. CosmicFish-HRM: Adaptive Reasoning via Hierarchical Recurrent Mechanisms in Compact Language Models

Authors: Venkat Akhil Lakkapragada
URL: https://arxiv.org/abs/2605.28919
Abstract:

Large language models have achieved strong reasoning capabilities, though often at the cost of massive parameter counts and expensive inference. In this work, we explore a different direction: adaptive reasoning depth in compact language models. We present CosmicFish-HRM, a compact language model built around a Hierarchical Reasoning Module (HRM) that dynamically allocates computational effort during inference. Instead of applying fixed computation to every input, the model iterates through high-level and low-level reasoning cycles and learns when to halt based on input complexity. CosmicFish-HRM combines this adaptive reasoning core with modern transformer components including Grouped Query Attention, RoPE, and SwiGLU activations. While the additional reasoning infrastructure introduces overhead at small scale, we hypothesize that this tradeoff becomes increasingly favorable as model size grows and the relative cost of the HRM core diminishes. Our results show that the model learns non-uniform reasoning behavior, allocating different numbers of reasoning steps across tasks and inputs. These findings suggest that adaptive reasoning depth may offer a promising alternative to relying solely on parameter scale for reasoning capability.

189. First head-to-head comparison of agentic AI applied to the analysis of simulated data of the Einstein Telescope

Authors: Gianluca Inguglia
URL: https://arxiv.org/abs/2605.28916
Abstract:

We report a comparison of two state-of-the-art agentic AI systems, Claude Code (Anthropic) and Codex (OpenAI), tasked with autonomously executing a simple end-to-end gravitational wave data analysis pipeline on a shared computing infrastructure without human intervention. The pipeline comprises power spectral density estimation from raw Einstein Telescope simulated noise, geometric template bank generation, matched filter recovery of 100 binary black hole signal injections, automated results generation, and large language model-assisted production of a manuscript formatted in the style of Physical Review D. Both agents received identical written specifications and identical compute resources. The experiment was run twice: a first run with unrealistically loud injections, and a second run with signals rescaled to a physically motivated SNR range. The scientific results converged in both runs. However, the agents exhibited substantially different behaviors and computational costs: Claude Code completed the pipeline in ~3.4 minutes with silent deviations from the specification, while Codex required ~16 minutes across explicit self-correcting restarts, including an unsolicited performance optimization of the matched filter inner loop. The autonomously generated manuscripts also diverged in length, details, and quality. In the second run, a subtle difference in the interpretation of the SNR range instruction led to a genuine scientific divergence: Claude Code silently reinterpreted the instructions, while Codex followed the specification literally. We discuss the implications of these behavioral differences, such as speed versus auditability, silent versus transparent error handling, instruction interpretation, and the criticality of intermediate data representations in multi-model pipelines, for the deployment of agentic AI in scientific computing workflows.

190. Hallucination Detection-Guided Preference Optimization for Clinical Summarization

Authors: Shamanth Kuthpadi Seethakantha , Dung Ngoc Thai , Vara Prasad Gudi , Simran Tiwari , Rami Matar , Avijit Mitra , Wenlong Zhao , Wael Salloum , Andrew McCallum
URL: https://arxiv.org/abs/2605.28910
Abstract:

Large language models (LLMs) have shown promise on summarization tasks, but they often produce hallucinations, which are unsupported or incorrect statements that limit their reliability in specialized healthcare applications. We introduce \itermodelfull (\itermodel), an inference-time method that leverages hallucination detectors to guide iterative summary revisions toward factual corrections. Building on this, we propose \itermodel for Preference Learning (\model), which converts detector-guided refinement trajectories into preference pairs for model finetuning. Extensive experiments show that our methods substantially reduce hallucinations for Llama and Gemma models in summarizing real-world clinical notes from \MimicIV. For example, \itermodel reduces 24\% and \model reduces 48\% hallucinations in Llama-3.1-8B-Instruct. Importantly, both methods preserve summary fluency, coherence, and relevance according to human expert and LLM-Jury evaluations. Together, these results demonstrate that detection-informed refinement and preference learning offer an automated solution for improving factual faithfulness in clinical summarization.

191. GrowLoop: Self-Evolving Conversation Evaluation Seeded by Human

Authors: Yihang Lin , Yunze Gao , Zeyang Lin , Dongbo Li , Kun Peng , Chenglong Song , Yue Liu
URL: https://arxiv.org/abs/2605.28882
Abstract:

With the rapid advancement of large language models, evaluating human-likeness in open-ended conversation has become increasingly important. However, human-likeness is a form of tacit knowledge that humans perceive intuitively, yet the underlying criteria resist explicit formulation. Human judgments vary widely, with strong agreement on some cases and legitimate disagreement on others. Meanwhile, the criteria behind human judgments remain implicit, leaving no clear basis for constructing cases. Further, what counts as human-like is not static, but evolving with model capability and human expectations. Despite progress in evaluation methods such as expert-authored benchmarks, Reward Models, and self-evolving benchmarks, none addresses all three challenges simultaneously. Therefore, we propose GrowLoop, a self-evolving conversation evaluation system that continuously adapts as models advance and scenarios shift. With minimal human seed annotations as the first mover, LLM agents iteratively extract and refine evaluation rubrics through Heuristic Learning. Human-AI agreement is required where annotators converge, while only plausibility is expected where they diverge. Moreover, the Rubric-Case co-evolution mechanism enables continuous evolution, expanded through new seeds when the evaluation target moves. Applied to human-likeness evaluation in open-ended conversation, the generated rubrics not only substantially outperform existing methods in alignment with human judgments, but also uncover issues that annotators overlook. The resulting benchmark effectively discriminates models across capability tiers and reveals where they fall short, while generalizing to new scenarios and adapting as models advance. Our work shifts the benchmarking paradigm from manual updates or difficulty scaling to comprehensive, continuous self-evolution.

192. LogDx-CI: Benchmarking Log Reduction Tools for LLM Root-Cause Diagnosis

Authors: Bowen Qin
URL: https://arxiv.org/abs/2605.28876
Abstract:

CI failure logs are large (median 5k lines, max 200k in this corpus) and noisy. Coding agents that try to debug them depend on an upstream tool to reduce the log to a manageable context, but the field has had no public empirical comparison of which reductions preserve enough evidence for downstream LLM diagnosis. We introduce LogDx-CI, a benchmark that compares 11 context-reduction tools (raw, tail, grep, three RTK modes, two real LLM map-reduce summarizers, three hybrid routers) on 35 real GitHub Actions failure cases, scored by 3 LLM debugger families (Claude Haiku 4.5, Claude Sonnet 4.6, OpenAI gpt-5-mini) plus a Sonnet 4.6 tool-using agent. We report three load-bearing findings. (1)~Hybrid grep+tail routers dominate the cost-quality Pareto frontier; the top two methods score 0.670 / 0.666 at $\sim$ $0.03 per case, same-ballpark quality as standalone grep at $4.5\times$ fewer tokens. (2)~In the agent-loop regime, the quality range across reduction tools collapses $7\times$ (single-shot spread 0.42 $\to$ agent-loop spread 0.059); the agent rescues weak contexts via follow-up tool calls. However, cost differences persist: weak contexts force the agent to issue 2–4$\times$ more tool calls to recover. (3)~A cross-family LLM-summary pair (gpt-5-mini summarizer feeding a Claude Haiku debugger) beats the same-family pair by $+0.071$ averaged across four diagnoser variants, falsifying the self-call-bias hypothesis on this task. The gpt-5-mini summarizer is also the agent-loop #1 method (score 0.749) at $0.37$ tool-calls per case and $10\times$ lower reducer cost than the Haiku summarizer ($0.18 vs $1.75 per case). All data, code, per-case bundles, and reproducibility infrastructure are public.

193. Continuity and Ordinality Matter: Constraining Time Series Tokens for Effective Time Series Analysis with Large Language Models

Authors: Musheng Li , Ziying Zhang , Cheng jin , Yuantao Gu
URL: https://arxiv.org/abs/2605.28866
Abstract:

Token-based time series large language models (TS-LLMs) have emerged as a promising direction for time series analysis and reasoning. However, prior studies largely overlook the inherent continuity and ordinality of time series tokens, which substantially limits model performance. In this paper, we argue that preserving these properties in time series token embeddings is crucial for the effectiveness of token-based TS-LLMs. To this end, we propose COM (Continuity and Ordinality Matter), a continuity- and ordinality-aware strategy that integrates geometric constraints into both the initialization and training stages. Empirical results on multiple time series analysis benchmarks demonstrate that COM consistently improves the performance of token-based TS-LLMs, achieving competitive results and strong generalizability. Code is available at this https URL .

194. Mechanistic origins of catastrophic forgetting: why RL preserves circuits better than SFT?

Authors: Jeanmely Rojas Nunez , Viraj Sawant , Nathan Allen , Nomgondalai Amgalanbaatar , Yannis Zongo , Vasu Sharma , Maheep Chaudhary
URL: https://arxiv.org/abs/2605.28860
Abstract:

Fine-tuning large language models (LLMs) frequently induces catastrophic forgetting of prior capabilities. Recent work has shown that reinforcement learning (RL) retains prior capabilities more effectively than supervised fine-tuning (SFT), attributing this to policy-gradient updates remaining closer to the base policy \cite{shenfeld2025rl}. We extend this behavioral account to the mechanistic level and ask whether RL’s advantage is mirrored by stronger preservation of internal computational circuits. We introduce differential circuit vulnerability, a head-level measure of how much a circuit degrades under fine-tuning, and use it to compare RL and SFT on Qwen2.5-3B-Instruct adapted to scientific question-answering. We find a clear mechanistic trade-off: SFT adapts more rapidly to the target task but produces substantially greater circuit disruption and forgetting of prior capabilities, whereas RL preserves a larger fraction of the base circuit at the cost of slower task adaptation. These findings suggest that circuit preservation may help explain why RL is more robust to catastrophic forgetting. We released our code here: this https URL .

195. GPF-LiveNews: A Streaming Evaluation Protocol for Group-Conditioned Framing in Large Language Models

Authors: Mohd Ariful Haque , Fahad Rahman , Kishor Datta Gupta , Roy George
URL: https://arxiv.org/abs/2605.28848
Abstract:

Deployed language models are evaluated in a non-stationary environment: model versions, retrieval layers, safety systems, and real-world inputs all change over time. Static bias benchmarks remain useful, but they do not show how models frame newly emerging events for different prompted audiences. We introduce GPF-LIVENEWS, a streaming evaluation protocol and benchmark snapshot for auditing group-conditioned framing in open-ended LLM outputs. The protocol expands fresh BBC/Reuters news anchors across 42 identity labels and seven prompt families, then evaluates response bundles using semantic-sensitivity and sentiment-disparity signals. In a pilot over 12 monitoring runs and 23 hosted models, Policy/Action prompts produce the strongest semantic movement, while sentiment variation is flatter across dimensions and prompt families. The released artifact includes article metadata, prompt templates, instantiated prompts, model-output metadata, score tables, documentation, and reproduction scripts. We interpret all scores as observed-window audit signals for human review, not as permanent fairness rankings or direct proof of harmful bias.

196. Thoughts-as-Planning: Latent World Models for Chain-of-Thoughts Optimization via Reinforcement Planning

Authors: Dong Liu , Yanxuan Yu , Ying Nian Wu
URL: https://arxiv.org/abs/2605.28842
Abstract:

The success of large language models (LLMs) across diverse NLP tasks has elevated the importance of reasoning chain optimization as a critical step in aligning model behavior with task objectives. Existing reasoning chain tuning methods often rely on black-box heuristics or gradient-free search, which lack interpretability, generalization, and sample efficiency. In this work, we introduce \textbf{Thoughts-as-Planning}, a novel framework that formalizes reasoning chain optimization as a sequential decision-making process over a latent semantic space. We model the LLM as a partially observable environment and learn a latent world model that simulates the effect of reasoning chain edits on downstream outputs. A proximity-preserving embedding space is constructed to encode reasoning chain-response dynamics, enabling planning via gradient descent or reinforcement learning. Our method supports multi-scale abstraction, allowing reasoning chain edits at token, segment, and instruction levels to be integrated into a unified planner. Through extensive experiments on language understanding and generation tasks, we demonstrate that Thoughts-as-Planning outperforms state-of-the-art reasoning chain tuning baselines in efficiency, robustness, and generalization, while offering interpretability through its structured planning trajectory. Our code is available at this https URL .

197. How Consistent Are LLM Agents? Measuring Behavioral Reproducibility in Multi-Step Tool-Calling Pipelines

Authors: Abel Yagubyan
URL: https://arxiv.org/abs/2605.28840
Abstract:

Large language model (LLM) agents with tool-calling capabilities are increasingly deployed in production systems, yet a fundamental reliability question remains under-explored: does the same agent behave the same way twice? We present a systematic empirical study of behavioral consistency in multi-step tool-calling agents, measuring whether agents select the same tools, in the same order, with the same arguments, across repeated identical invocations. Unlike prior work on consistency in ReAct-style agents(search-only, free-text actions), we study the richer setting of structured tool-calling interfaces with typed parameters and consequential side effects.

198. Specialty-Specific Medical Language Model for Immune-Mediated Diseases

Authors: Veysel Kocaman , Gursev Pirge , Yigit Gul , Ace Vo , Zhenya Nargizyan , David Talby
URL: https://arxiv.org/abs/2605.28838
Abstract:

Extracting detailed clinical information from free-text medical narratives remains a practical challenge for researchers and healthcare systems. Terminology for immune-mediated and infectious diseases is especially inconsistent across sources, which often limits the ability of general-purpose Natural Language Processing (NLP) systems to capture the relevant biomedical concepts with sufficient granularity. We developed a domain-specific Named Entity Recognition (NER) model tailored to identify disease-related entities occurring in immunology and infectious disease contexts. We assembled and manually annotated a dataset of 371 case reports in collaboration with two clinical specialists, defining twelve entity classes covering immune-mediated and infectious conditions as well as related symptoms and clinical descriptors. We evaluated several modeling strategies, including the MedicalNER architecture with multiple healthcare-specific embeddings, a BERT-based token classification model, and zero-shot NER systems. The strongest performance was obtained with a transformer-based model trained on clinical-domain embeddings, which reached an F1 score of 0.89, consistently outperforming baseline and zero-shot approaches. The combination of specialized embeddings and expert annotation proved particularly valuable for capturing nuanced disease terminology and improving generalization across heterogeneous biomedical text. The prompted LLM baseline achieved substantially lower performance under the same evaluation protocol, reflecting difficulties in producing span-consistent outputs for fine-grained entity boundaries despite detailed prompting. The resulting model provides a structured way to analyze case reports and can support downstream tasks such as cohort identification, disease monitoring, and clinical decision support.

199. SERC: LDPC-Inspired Semantic Error Correction for Retrieval-Augmented Generation

Authors: Gyumin Kim , Juhwan Park , Jaeha Kim , Seunggyun Han , Kyungrak Son , Ikbeom Jang
URL: https://arxiv.org/abs/2605.28837
Abstract:

While Large Language Models (LLMs) have demonstrated remarkable capabilities, their reliability is significantly compromised by hallucinations. Existing intrinsic self-correction methods attempt to address this, but often fail due to self-bias, where models struggle to identify errors in their own outputs without external verification. To overcome these limitations, we propose the LDPC-inspired semantic error correction for retrieval-augmented generation (SERC), providing a theoretical framework to interpret and mitigate LLM hallucinations. We reformulate the text generation process as a semantic noisy channel, treating generated responses as noise-corrupted codewords. Inspired by low-density parity-check (LDPC) codes, SERC employs a sparse verification strategy: instead of exhaustively checking all facts, it generates low-density verification queries and validates them against external evidence to efficiently detect and correct errors. We evaluate SERC on LongForm Bio and TruthfulQA benchmarks using Llama-3-8B and Qwen2.5-14B. Experimental results demonstrate that SERC outperforms both intrinsic self-correction methods and strong retrieval-augmented baselines, demonstrating significant gains especially in factual precision (FactScore). Notably, SERC enables small language models (SLMs) to surpass the performance of larger baselines in hallucination reduction and information preservation. Our findings demonstrate that SERC provides a training-free, model-agnostic solution that significantly reduces verification overhead compared to dense methods, achieving an optimal trade-off between cost and fidelity in resource-constrained environments.

200. GenesisFunc: Multi-Agent Data Generation for Accurate and Generalizable Function-Calling

Authors: Hao-Xiang Xu , Chong Deng , Jiaqing Liu , Wen Wang , Qian Chen , Lujia Bao , Xiangang Li , Zhen-Hua Ling
URL: https://arxiv.org/abs/2605.28835
Abstract:

Large Language Models (LLMs) extend their capabilities through function-calling (FC), which relies on training data with high quality, diversity, and broad coverage of scenario. However, obtaining and annotating real function-calling data is challenging, while synthetic data from existing pipelines often suffers from unreliable APIs, limited tool scalability, insufficient diversity, and weak quality control. To address these, we present GenesisFunc, an automated pipeline for generating FC training data. Starting from reliable tools in widely used public benchmarks, our GenesisFunc employs a multi-agent framework to support a dialogue generation system that produces conversations spanning diverse scenarios, while maintaining both diversity and quality throughout the process. The accuracy of the data is further reinforced through a multi-stage evaluation system. We fine-tune an 8B LLM on the synthetic dataset and show through extensive experiments that it outperforms similarly sized open-source models in in-domain FC performance and out-of-domain generalization, while reaching FC capabilities comparable to some of the latest API-based models. In addition, our method demonstrates strong potential to scale effectively across downstream tools, underscoring its real-world applicability.

201. A comparative study of transformer-based embeddings for topic coherence

Authors: Alex Ding , Tarun Rapaka , Willy Rodriguez , Jason Yang
URL: https://arxiv.org/abs/2605.28832
Abstract:

Topic modeling is a branch of Natural Language Processing (NLP) that aims to organize large collections of texts into coherent groups according to word co-occurrence patterns, with Latent Dirichlet Allocation (LDA) remaining one of the most widely used and interpretable probabilistic approaches. Recent advances in NLP, particularly transformer-based language models, offer improved document representations. It is also known that the size of the model (in terms of number of parameters) has a significant impact in the performance of the language models on different pre-defined tasks. In this study, we systematically examine the effect of model size on topic quality by analyzing the performances of seven transformer-based language models (from small models such as MiniLM to large ones such as LLaMA-2) in a BERTopic pipeline on a variety of corpora. Topic quality is evaluated using coherence and divergence metrics following R{ö}der et al. (2015). Our results indicate that model size, ranging from 22 million to 13 billion parameters, has a negligible impact on the quality of the topic, suggesting that smaller models can achieve comparable performance to larger models.

202. Benchmarking Open-Source Safety Guard Models: A Comprehensive Evaluation

Authors: Reetu Raj Harsh , Bhaskarjit Sarmah , Stefano Pasquali
URL: https://arxiv.org/abs/2605.28830
Abstract:

As Large Language Models (LLMs) are increasingly deployed in safety-critical applications, robust content moderation becomes essential. We present a comprehensive evaluation of 14 open-source safety guard models on a curated benchmark of 79,331 samples spanning 8 NIST AI Risk Framework safety categories. Our benchmark aggregates four diverse datasets (HarmBench, StrongREJECT, RealToxicityPrompts, and BeaverTails), filtered to focus exclusively on safety-relevant content (violence, hate speech, harassment, sexual content, suicide/self-harm, profanity, threats, and health misinformation). We find that recall is the critical metric for safety applications, as missing harmful content poses greater risk than false positives. Our evaluation reveals surprising results: Qwen Guard (4B parameters) achieves the highest recall (83.97%) while larger models like Llama Guard (12B) and GPT-OSS Safeguard (20B) exhibit conservative behavior, missing up to 75% of unsafe content. We demonstrate that model size does not correlate with safety detection performance and that general-purpose guard models outperform specialized ones. These findings provide practical guidance for selecting safety guard models in production deployments.

203. Aryabhata 2: Scaling Reinforcement Learning for Advanced STEM Reasoning

Authors: Ritvik Rastogi , Vishal Singh , Tejas Chaudhari , Sandeep Varma
URL: https://arxiv.org/abs/2605.28829
Abstract:

Competitive STEM examinations such as JEE and NEET require multi-step symbolic reasoning, precise numerical computation, and deep conceptual understanding across physics, chemistry, and mathematics. Recent large language models perform strongly on common reasoning benchmarks, yet they remain difficult to deploy at scale, where millions of student doubts demand domain-specific, consistently structured problem solving. We introduce Aryabhata 2, a reasoning-focused language model for competitive STEM examinations, trained via reinforcement-learning post-training. Using PhysicsWallah’s internal question banks, we construct a high-quality training curriculum and post-train GPT-OSS-20B through reinforcement learning with verifiable rewards. Training combines prolonged reinforcement learning with broadened exploration via progressively larger rollout group sizes. We evaluate Aryabhata 2 on competitive examination benchmarks, including JEE Main, JEE Advanced, and NEET, as well as out-of-distribution reasoning datasets such as AIME, HMMT, MMLU-Pro, MMLU-Redux 2.0, and GPQA. Results show that Aryabhata 2 outperforms its base model GPT-OSS-20B on competitive STEM reasoning while requiring substantially fewer output tokens (up to 64\% fewer).

204. Micro-Macro Retrieval: Reducing Long-Form Hallucination in Large Language Models

Authors: Yujie Feng , Jian Li , Zhihan Zhou , Pengfei Xu , Yujia Zhang , Xiaoyu Li , Xiaohui Zhou , Alan Zhao , Xi Chen , Xiao-Ming Wu
URL: https://arxiv.org/abs/2605.28828
Abstract:

Large Language Models (LLMs) achieve impressive performance across many tasks but remain prone to hallucination, especially in long-form generation where redundant retrieved contexts and lengthy reasoning chains amplify factual errors. Recent studies highlight a critical phenomenon: the closer key information appears to the model outputs, the higher the factual accuracy. However, existing retrieval-augmented language models (RALMs) lack effective mechanisms to ensure this proximity - external evidence is injected into reasoning via multi-turn retrieval, but this cannot ensure key information stays close to the outputs. We propose Micro-Macro Retrieval (M2R), a novel retrieve-while-generate framework to fill this gap. At the macro level, M2R retrieves coarse-grained evidence from external sources; at the micro level, it extracts essential results from a key information repository built during reasoning and reuses them while generating answers. This design directly addresses the key-information-to-output proximity bottleneck, effectively reducing hallucination in long-form tasks. M2R is trained with a curriculum learning-based reinforcement learning strategy using customized rule-based rewards, enabling stable acquisition of retrieval and grounding skills. Extensive experiments across different benchmarks demonstrate the effectiveness of M2R, especially in lengthy-context settings.

LLM 관련 주요 논문 - 2026-05-29

1. SchGen: PCB Schematic Generation with Semantic-Grounded Code Representations

2. Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection

3. Locally Coherent, Globally Incoherent: Bounding Compositional Incoherence in Multi-Component LLM Agents

4. Demystifying Data Organization for Enhanced LLM Training

5. MIRA: Mid-training Rubric Anchoring for Source-Aware Data Selection

6. ProjectionBench: Evaluating Scientific Hypothesis Generation in LLMs Under Progressive Information Disclosure

7. When Should Models Change Their Minds? Contextual Belief Management in Large Language Models

8. Double-Edged Sword or Sharp Tool? Designing and Evaluating Triadic LLM-Teacher Collaboration for K-12 Writing at Scale

9. Modularizing Educational LLM-Agency for Fostering Responsible Learning Assistance

10. BioRefusalAudit: Auditing Biosecurity Refusal Depth Using General and Domain-Fine-Tuned Sparse Autoencoders

11. Meta-Cognitive Memory Policy Optimization for Long-Horizon LLM Agents

12. Anchorless Diversification for Parallel LLM Ideation

13. AgentSchool: An LLM-Powered Multi-Agent Simulation for Education

14. Enhancing Multi-Agent Communication through Attention Steering with Context Relevance

15. PokerSkill: LLMs Can Play Expert-Level Poker without Training or Solvers

16. Selective QA over Conflicting Multi-Source Personal Memory: A Diagnostic Testbed and Method Comparison

17. Conformal Certification of Reasoning Trace Prefixes

18. Learning to Choose: An Empowerment-Guided Multi-Agent System with semantic communication for Adaptive Method Selection

19. Domain-Specific Data Synthesis for LLMs via Minimal Sufficient Representation Learning

20. Teaching Values to Machines: Simulating Human-Like Behavior in LLMs

21. From GPS Points to Travel Patterns: Flexible and Semantic Trajectory Generation with LLMs

22. KairosAgent: Agentic Time Series Forecasting with Fused Semantic Reasoning

23. Cookie-Bench: Continuous On-screen Key Interaction Evaluation for Web Generation

24. Accelerating Constrained Decoding with Token Space Compression

25. Compass: Navigating Global Marine Lead Data Integration through Expert-Guided LLM Agent

26. Formalizing Mathematics at Scale

27. MuPHI: Learning Implicit Multimodal Harm Reasoning via Semantically Grounded Reward Optimization

28. Make LLM Learn to Synthesize from Streaming Experiences through Feedback

29. Redundant or Necessary? A Benchmark for Detecting Redundant Steps in Agent Trajectories

30. Moment-KV: Momentum-Based Decode-Time KV Cache Compression for Long Generation

31. OmniMatBench: A Human-Calibrated Multimodal Reasoning Benchmark Across 19 Materials Science Subfields

32. OptSkills: Learning Generalizable Optimization Skills from Problem Archetypes via Cluster-Based Distillation

33. Quantifying and Optimizing Simplicity via Polynomial Representations

34. Harnessing non-adversarial robustness in large language models

35. PRAIB: Peer Review AI Benchmark of Behaviour of LLM-Assisted Reviewing

36. SkillsInjector: Dynamic Skill Context Construction for LLM Agents

37. Croissant Tasks: A Metadata Format for Reproducible Machine Learning Evaluations

38. LFQ: Logit-aware Final-block Quantization for Boosting the Generation Quality of Low-Bit Quantized LLMs

39. Why Specialist Models Still Matter: A Heterogeneous Multi-Agent Paradigm for Medical Artificial Intelligence

40. Citation-Closure Retrieval and Per-Rule Attribution for Real-World Regulatory Compliance Question Answering

41. NaRA: Noise-Aware LoRA for Parameter-Efficient Fine-Tuning of Diffusion LLMs

42. BitTP: The Lightweight Trajectory Prediction Model with BitLLM for Edge-Devices

43. Reliable Reasoning with Large Language Models via Preference-Based Maximum Satisfiability

44. NICE: A Theory-Grounded Diagnostic Benchmark for Social Intelligence of LLMs

45. Notation Matters: A Benchmark Study of Token-Optimized Formats in Agentic AI Systems

46. GRASP: Gated Regression-Aware Skill Proposer for Self-Improving LLM Agents

47. TRACE: Toulmin-based Reasoning Assessment through Constructive Elements for LLM CoT Evaluation

48. PTCG-Bench: Can LLM Agents Master Pokémon Trading Card Game?

49. Think Fast, Talk Smart: Partitioning Deterministic and Neural Computation for Structured Health Text Generation

50. LLM-Evolved Domain-Independent Heuristics for Symbolic AI Planning

51. VikingMem: A Memory Base Management System for Stateful LLM-based Applications

52. Beyond Attack Success Rate: Temporal Logit Observability for LLM Safety Failures

53. Improving Collaborative Storytelling with a Multi-Agent Framework Based on Large Language Models

54. FinVerBench: Benchmark Validity and Calibration in Large Language Model Financial Statement Verification

55. GPS-Enhanced Tourist Mobility Modeling with Seasonal Spatial Priors and LLM-Based Activity Chain Generation

56. DeepTool: Scaling Interleaved Deliberation in Tool-Integrated Reasoning via Process-Supervised Reinforcement Learning

57. ParaTool: Shifting Tool Representations from Context to Parameters

58. Battery-Sim-Agent: Leveraging LLM-Agent for Inverse Battery Parameter Estimation

59. Opt-Verifier: Unleashing the Power of LLMs for Optimization Modeling via Dual-Side Verification

60. UI-KOBE: Knowledge-Oriented Behavior Exploration for Lightweight Graph-Guided GUI Agents

61. MINDGAMES: A Live Arena for Evaluating Social and Strategic Reasoning in Multi-Agent LLMs

62. The Curse of Helpfulness: Inverse Scaling Law in Robustness to Distractor Instructions via DistractionIF

63. CrystalXRD-Bench: Benchmarking Vision-Language Models for XRD Peak Indexing Across Diverse Crystalline Materials

64. Towards Human-Like Interactive Speech Recognition With Agentic Correction and Semantic Evaluation

65. When Does Persona Prompting Actually Help? A Retrieval and Metric Analysis of Expert Role Injection in LLMs

66. Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization

67. EvoMD-LLM: Learning the Language of Species Evolution in Reactive Molecular Dynamics

68. Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet

69. PassNet: Scaling Large Language Models for Graph Compiler Pass Generation

70. ConMoE: Expert-Pool Consolidation via Prototype Reassignment for MoE Compression

71. Entropy-KL Divergence-based Token Masking: A Novel Approach for Selective Fine-tuning of Large Language Models

72. Diagnosing Harmful Continuation in Answer-Correct Long-CoT Training Traces

73. CoHyDE: Iterative Co-Training of LLM Rewriter & Dense Encoder for Tool Retrieval

74. Indexing the Unreadable: LLM-Native Recursive Construction and Search of Service Taxonomies

75. Harmonizing Real-Time Constraints and Long-Horizon Reasoning: An Asynchronous Agentic Framework for Dynamic Scheduling

76. Provably Secure Agent Guardrail

77. DenseSteer: Steering Small Language Models towards Dense Math Reasoning

78. Rethinking Literature Search Evaluation: Deep Research Helps, and Human Citation Lists Are Not a Ground Truth

79. Tailoring the Curriculum: Student-Centered Reasoning Distillation via Dynamic Data-Model Compatibility