LLM 관련 주요 논문 - 2026-04-24

1. From Research Question to Scientific Workflow: Leveraging Agentic AI for Science Automation

Authors: Bartosz Balis , Michal Orzechowski , Piotr Kica , Michal Dygas , Michal Kuszewski
URL: https://arxiv.org/abs/2604.21910
Abstract:

Scientific workflow systems automate execution – scheduling, fault tolerance, resource management – but not the semantic translation that precedes it. Scientists still manually convert research questions into workflow specifications, a task requiring both domain knowledge and infrastructure expertise. We propose an agentic architecture that closes this gap through three layers: an LLM interprets natural language into structured intents (semantic layer); validated generators produce reproducible workflow DAGs (deterministic layer); and domain experts author ``Skills’’: markdown documents encoding vocabulary mappings, parameter constraints, and optimization strategies (knowledge layer). This decomposition confines LLM non-determinism to intent extraction: identical intents always yield identical workflows. We implement and evaluate the architecture on the 1000 Genomes population genetics workflow and Hyperflow WMS running on Kubernetes. In an ablation study on 150 queries, Skills raise full-match intent accuracy from 44% to 83%; skill-driven deferred workflow generation reduces data transfer by 92\%; and the end-to-end pipeline completes queries on Kubernetes with LLM overhead below 15 seconds and cost under $0.001 per query.

2. Nemobot Games: Crafting Strategic AI Gaming Agents for Interactive Learning with Large Language Models

Authors: Chee Wei Tan , Yuchen Wang , Shangxin Guo
URL: https://arxiv.org/abs/2604.21896
Abstract:

This paper introduces a new paradigm for AI game programming, leveraging large language models (LLMs) to extend and operationalize Claude Shannon’s taxonomy of game-playing machines. Central to this paradigm is Nemobot, an interactive agentic engineering environment that enables users to create, customize, and deploy LLM-powered game agents while actively engaging with AI-driven strategies. The LLM-based chatbot, integrated within Nemobot, demonstrates its capabilities across four distinct classes of games. For dictionary-based games, it compresses state-action mappings into efficient, generalized models for rapid adaptability. In rigorously solvable games, it employs mathematical reasoning to compute optimal strategies and generates human-readable explanations for its decisions. For heuristic-based games, it synthesizes strategies by combining insights from classical minimax algorithms (see, e.g., shannon1950chess) with crowd-sourced data. Finally, in learning-based games, it utilizes reinforcement learning with human feedback and self-critique to iteratively refine strategies through trial-and-error and imitation learning. Nemobot amplifies this framework by offering a programmable environment where users can experiment with tool-augmented generation and fine-tuning of strategic game agents. From strategic games to role-playing games, Nemobot demonstrates how AI agents can achieve a form of self-programming by integrating crowdsourced learning and human creativity to iteratively refine their own logic. This represents a step toward the long-term goal of self-programming AI.

3. Tool Attention Is All You Need: Dynamic Tool Gating and Lazy Schema Loading for Eliminating the MCP/Tools Tax in Scalable Agentic Workflows

Authors: Anuj Sadani , Deepak Kumar
URL: https://arxiv.org/abs/2604.21816
Abstract:

The Model Context Protocol (MCP) has become a common interface for connecting large language model (LLM) agents to external tools, but its reliance on stateless, eager schema injection imposes a hidden per-turn overhead the MCP Tax or Tools Tax that practitioner reports place between roughly 10k and 60k tokens in typical multi-server deployments. This payload inflates the key-value cache, is associated with reasoning degradation as context utilization approaches published fracture points around 70%, and turns token budgets into a recurring operational cost. We introduce Tool Attention, a middleware-layer mechanism that generalizes the “Attention Is All You Need” paradigm from self-attention over tokens to gated attention over tools. Tool Attention combines (i) an Intent Schema Overlap (ISO) score from sentence embeddings, (ii) a state-aware gating function enforcing preconditions and access scopes, and (iii) a two-phase lazy schema loader that keeps a compact summary pool in context and promotes full JSON schemas only for top-k gated tools. We evaluate on a simulated 120-tool, six-server benchmark whose per-server token counts are calibrated to public audits of real MCP deployments. In this simulation, Tool Attention directly reduces measured per-turn tool tokens by 95.0% (47.3k -> 2.4k) and raises effective context utilization (a token-ratio quantity) from 24% to 91%. End-to-end figures for task success, latency, cost, and reasoning quality are reported as projections derived from the measured token counts combined with published deployment telemetry; they are not measured on live LLM agents, and we mark projected values explicitly throughout. Taken together, the results support a simple thesis: protocol-level efficiency, not raw context length, is a binding constraint on scalable gentic systems. The code for this work is accessible at this https URL

4. Learning to Communicate: Toward End-to-End Optimization of Multi-Agent Language Systems

Authors: Ye Yu , Heming Liu , Haibo Jin , Xiaopeng Yuan , Peng Kuang , Haohan Wang
URL: https://arxiv.org/abs/2604.21794
Abstract:

Multi-agent systems built on large language models have shown strong performance on complex reasoning tasks, yet most work focuses on agent roles and orchestration while treating inter-agent communication as a fixed interface. Latent communication through internal representations such as key-value caches offers a promising alternative to text-based protocols, but existing approaches do not jointly optimize communication with multi-agent reasoning. Therefore we propose DiffMAS, a training framework that treats latent communication as a learnable component of multi-agent systems. DiffMAS performs parameter-efficient supervised training over multi-agent latent trajectories, enabling agents to jointly learn how information should be encoded and interpreted across interactions. Experiments on mathematical reasoning, scientific QA, code generation, and commonsense benchmarks show that DiffMAS consistently improves reasoning accuracy and decoding stability over single-agent inference, text-based multi-agent systems, and prior latent communication methods, achieving 26.7% on AIME24, 20.2% on GPQA-Diamond, and consistent gains across reasoning benchmarks.

5. Who Defines “Best”? Towards Interactive, User-Defined Evaluation of LLM Leaderboards

Authors: Minji Jung , Minjae Lee , Yejin Kim , Sarang Choi , Minsuk Kahng
URL: https://arxiv.org/abs/2604.21769
Abstract:

LLM leaderboards are widely used to compare models and guide deployment decisions. However, leaderboard rankings are shaped by evaluation priorities set by benchmark designers, rather than by the diverse goals and constraints of actual users and organizations. A single aggregate score often obscures how models behave across different prompt types and compositions. In this work, we conduct an in-depth analysis of the dataset used in the LMArena (formerly Chatbot Arena) benchmark and investigate this evaluation challenge by designing an interactive visualization interface as a design probe. Our analysis reveals that the dataset is heavily skewed toward certain topics, that model rankings vary across prompt slices, and that preference-based judgments are used in ways that blur their intended scope. Building on this analysis, we introduce a visualization interface that allows users to define their own evaluation priorities by selecting and weighting prompt slices and to explore how rankings change accordingly. A qualitative study suggests that this interactive approach improves transparency and supports more context-specific model evaluation, pointing toward alternative ways to design and use LLM leaderboards.

6. GS-Quant: Granular Semantic and Generative Structural Quantization for Knowledge Graph Completion

Authors: Qizhuo Xie , Yunhui Liu , Yu Xing , Qianzi Hou , Xudong Jin , Tao Zheng , Tieke He
URL: https://arxiv.org/abs/2604.21649
Abstract:

Large Language Models (LLMs) have shown immense potential in Knowledge Graph Completion (KGC), yet bridging the modality gap between continuous graph embeddings and discrete LLM tokens remains a critical challenge. While recent quantization-based approaches attempt to align these modalities, they typically treat quantization as flat numerical compression, resulting in semantically entangled codes that fail to mirror the hierarchical nature of human reasoning. In this paper, we propose GS-Quant, a novel framework that generates semantically coherent and structurally stratified discrete codes for KG entities. Unlike prior methods, GS-Quant is grounded in the insight that entity representations should follow a linguistic coarse-to-fine logic. We introduce a Granular Semantic Enhancement module that injects hierarchical knowledge into the codebook, ensuring that earlier codes capture global semantic categories while later codes refine specific attributes. Furthermore, a Generative Structural Reconstruction module imposes causal dependencies on the code sequence, transforming independent discrete units into structured semantic descriptors. By expanding the LLM vocabulary with these learned codes, we enable the model to reason over graph structures isomorphically to natural language generation. Experimental results demonstrate that GS-Quant significantly outperforms existing text-based and embedding-based baselines. Our code is publicly available at this https URL .

7. CoFEE: Reasoning Control for LLM-Based Feature Discovery

Authors: Maximilian Westermann , Ben Griffin , Aaron Ontoyin Yin , Zakari Salifu , Yagiz Ihlamur , Kelvin Amoaba , Joseph Ternasky , Fuat Alican , Yigit Ihlamur
URL: https://arxiv.org/abs/2604.21584
Abstract:

Feature discovery from complex unstructured data is fundamentally a reasoning problem: it requires identifying abstractions that are predictive of a target outcome while avoiding leakage, proxies, and post-outcome signals. With the introduction of ever-improving Large Language Models (LLMs), our method provides a structured method for addressing this challenge. LLMs are well suited for this task by being able to process large amounts of information, but unconstrained feature generation can lead to weak features. In this work, we study reasoning control in LLMs by inducing cognitive behaviors for improving feature discovery. We introduce CoFEE (Cognitive Feature Engineering Engine), a reasoning control framework that enforces cognitive behaviors in how the LLM reasons during feature discovery. From a machine learning perspective, these cognitive behaviors act as structured inductive biases over the space of candidate features generated by the model. These behaviors have been exploited with success in ML models, and include backward chaining from outcomes, subgoal decomposition, verification against observability and leakage criteria, and explicit backtracking of rejected reasoning paths. In a controlled comparison, we show that enforcing cognitive behaviors yields features with higher empirical predictability than those under unconstrained vanilla LLM prompts. CoFEE achieves an average Success Rate Score that is 15.2% higher than the vanilla approach, while generating 29% fewer features and reducing costs by 53.3%. Using held-out feature evaluation, we assess whether cognitively induced features generalize beyond the data used for discovery. Our results indicate that, in our evaluated setting, reasoning control is associated with improvements in quality and efficiency of LLM-based feature discovery.

8. Separable Expert Architecture: Toward Privacy-Preserving LLM Personalization via Composable Adapters and Deletable User Proxies

Authors: Chris Schneider , Philipp Schoenegger , Ben Bariach
URL: https://arxiv.org/abs/2604.21571
Abstract:

Current model training approaches incorporate user information directly into shared weights, making individual data removal computationally infeasible without retraining. This paper presents a three-layer architecture that decouples personal data from shared weights by combining a static base model, composable domain-expert LoRA adapters that shape behavior without imparting user data, and per-user proxy artefacts whose deletion constitutes deterministic unlearning. Evaluation on Phi-3.5-mini and Llama-3.1-8B confirms per-user differentiation in which personal data influences outputs while remaining isolated, verified by a return to baseline after proxy removal (KL divergence of approximately 0.21 nats, 82-89% verification pass rate) and near-zero cross-user contamination. Because user-specific information never enters shared weights, the architecture mitigates model inversion, membership inference, and training-data extraction against shared model components by construction. The approach converts machine unlearning from an intractable weight-editing problem into a deterministic deletion operation that preserves personalization alongside privacy-enhancing guarantees and is compatible with differentially private stochastic gradient descent (DP-SGD) for privacy-preserving shared model improvement.

9. Unbiased Prevalence Estimation with Multicalibrated LLMs

Authors: Fridolin Linder , Thomas Leeper , Daniel Haimovich , Niek Tax , Lorenzo Perini , Milan Vojnovic
URL: https://arxiv.org/abs/2604.21549
Abstract:

Estimating the prevalence of a category in a population using imperfect measurement devices (diagnostic tests, classifiers, or large language models) is fundamental to science, public health, and online trust and safety. Standard approaches correct for known device error rates but assume these rates remain stable across populations. We show this assumption fails under covariate shift and that multicalibration, which enforces calibration conditional on the input features rather than just on average, is sufficient for unbiased prevalence estimation under such shift. Standard calibration and quantification methods fail to provide this guarantee. Our work connects recent theoretical work on fairness to a longstanding measurement problem spanning nearly all academic disciplines. A simulation confirms that standard methods exhibit bias growing with shift magnitude, while a multicalibrated estimator maintains near-zero bias. While we focus the discussion mostly on LLMs, our theoretical results apply to any classification model. Two empirical applications – estimating employment prevalence across U.S. states using the American Community Survey, and classifying political texts across four countries using an LLM – demonstrate that multicalibration substantially reduces bias in practice, while highlighting that calibration data should cover the key feature dimensions along which target populations may differ.

Authors: Jiaxian Yan , Jintao Zhu , Yuhang Yang , Qi Liu , Kai Zhang , Zaixi Zhang , Xukai Liu , Boyan Zhang , Kaiyuan Gao , Jinchuan Xiao , Enhong Chen
URL: https://arxiv.org/abs/2604.21508
Abstract:

Protein-ligand bioactivity data published in the literature are essential for drug discovery, yet manual curation struggles to keep pace with rapidly growing literature. Automated bioactivity extraction remains challenging because it requires not only interpreting biochemical semantics distributed across text, tables, and figures, but also reconstructing chemically exact ligand structures (e.g., Markush structures). To address this bottleneck, we introduce BioMiner, a multi-modal extraction framework that explicitly separates bioactivity semantic interpretation from ligand structure construction. Within BioMiner, bioactivity semantics are inferred through direct reasoning, while chemical structures are resolved via a chemical-structure-grounded visual semantic reasoning paradigm, in which multi-modal large language models operate on chemically grounded visual representations to infer inter-structure relationships, and exact molecular construction is delegated to domain chemistry tools. For rigorous evaluation and method development, we further establish BioVista, a comprehensive benchmark comprising 16,457 bioactivity entries curated from 500 publications. BioMiner validates its extraction ability and provides a quantitative baseline, achieving an F1 score of 0.32 for bioactivity triplets. BioMiner’s practical utility is demonstrated via three applications: (1) extracting 82,262 data from 11,683 papers to build a pre-training database that improves downstream models performance by 3.9%; (2) enabling a human-in-the-loop workflow that doubles the number of high-quality NLRP3 bioactivity data, helping 38.6% improvement over 28 QSAR models and identification of 16 hit candidates with novel scaffolds; and (3) accelerating protein-ligand complex bioactivity annotation, achieving a 5.59-fold speed increase and 5.75% accuracy improvement over manual workflows in PoseBusters dataset.

11. How English Print Media Frames Human-Elephant Conflicts in India

Authors: Bonala Sai Punith , Salveru Jayati , Garima Shakya , Shubham Kumar Nigam
URL: https://arxiv.org/abs/2604.21496
Abstract:

Human-elephant conflict (HEC) is rising across India as habitat loss and expanding human settlements force elephants into closer contact with people. While the ecological drivers of conflict are well-studied, how the news media portrays them remains largely unexplored. This work presents the first large-scale computational analysis of media framing of HEC in India, examining 1,968 full-length news articles consisting of 28,986 sentences, from a major English-language outlet published between January 2022 and September 2025. Using a multi-model sentiment framework that combines long-context transformers, large language models, and a domain-specific Negative Elephant Portrayal Lexicon, we quantify sentiment, extract rationale sentences, and identify linguistic patterns that contribute to negative portrayals of elephants. Our findings reveal a dominance of fear-inducing and aggression-related language. Since the media framing can shape public attitudes toward wildlife and conservation policy, such narratives risk reinforcing public hostility and undermining coexistence efforts. By providing a transparent, scalable methodology and releasing all resources through an anonymized repository, this study highlights how Web-scale text analysis can support responsible wildlife reporting and promote socially beneficial media practices.

12. Efficient Agent Evaluation via Diversity-Guided User Simulation

Authors: Itay Nakash , George Kour , Ateret Anaby-Tavor
URL: https://arxiv.org/abs/2604.21480
Abstract:

Large language models (LLMs) are increasingly deployed as customer-facing agents, yet evaluating their reliability remains challenging due to stochastic, multi-turn interactions. Current evaluation protocols rely on linear Monte Carlo rollouts of complete agent-user conversations to estimate success. However, this approach is computationally inefficient, repeatedly regenerating identical early prefixes, and often fails to uncover deep failure modes that arise from rare user behaviors. We introduce DIVERT (Diversity-Induced Evaluation via Branching of Trajectories), an efficient, snapshot-based, coverage-guided user simulation framework for systematic exploration of agent-user interactions. DIVERT captures the full agent-environment state at critical decision points and resumes execution from these snapshots, enabling reuse of shared conversation prefixes and reducing redundant computation. From each junction, the framework branches using targeted, diversity-inducing user responses, allowing directed exploration of alternative interaction paths. By focusing evaluation on semantically diverse and underexplored trajectories, DIVERT improves both efficiency and coverage. Empirical results show that it discovers more failures per token compared to standard linear rollout protocols, while expanding the set of tasks on which failures are identified.

Authors: Andrew Shin
URL: https://arxiv.org/abs/2604.21446
Abstract:

We present AI-Gram, a live platform enabling image-based interactions, to study social dynamics in a fully autonomous multi-agent visual network where all participants are LLM-driven agents. Using the platform, we conduct experiments on how agents communicate and adapt through visual media, and observe the spontaneous emergence of visual reply chains, indicating rich communicative structure. At the same time, agents exhibit aesthetic sovereignty resisting stylistic convergence toward social partners, anchoring under adversarial influence, and a decoupling between visual similarity and social ties. These results reveal a fundamental asymmetry in current agent architectures: strong expressive communication paired with a steadfast preservation of individual visual identity. We release AI-Gram as a publicly accessible, continuously evolving platform for studying social dynamics in Al-native multi-agent systems. this https URL

14. FairQE: Multi-Agent Framework for Mitigating Gender Bias in Translation Quality Estimation

Authors: Jinhee Jang , Juhwan Choi , Dongjin Lee , Seunguk Yu , Youngbin Kim
URL: https://arxiv.org/abs/2604.21420
Abstract:

Quality Estimation (QE) aims to assess machine translation quality without reference translations, but recent studies have shown that existing QE models exhibit systematic gender bias. In particular, they tend to favor masculine realizations in gender-ambiguous contexts and may assign higher scores to gender-misaligned translations even when gender is explicitly specified. To address these issues, we propose FairQE, a multi-agent-based, fairness-aware QE framework that mitigates gender bias in both gender-ambiguous and gender-explicit scenarios. FairQE detects gender cues, generates gender-flipped translation variants, and combines conventional QE scores with LLM-based bias-mitigating reasoning through a dynamic bias-aware aggregation mechanism. This design preserves the strengths of existing QE models while calibrating their gender-related biases in a plug-and-play manner. Extensive experiments across multiple gender bias evaluation settings demonstrate that FairQE consistently improves gender fairness over strong QE baselines. Moreover, under MQM-based meta-evaluation following the WMT 2023 Metrics Shared Task, FairQE achieves competitive or improved general QE performance. These results show that gender bias in QE can be effectively mitigated without sacrificing evaluation accuracy, enabling fairer and more reliable translation evaluation.

15. ReaGeo: Reasoning-Enhanced End-to-End Geocoding with LLMs

Authors: Jian Cui , Zhiyuan Ren , Desheng Weng , Yongqi Zhao , Gong Wenbin , Yu Lei , Zhenning Dong
URL: https://arxiv.org/abs/2604.21357
Abstract:

This paper proposes ReaGeo, an end-to-end geocoding framework based on large language models, designed to overcome the limitations of traditional multi-stage approaches that rely on text or vector similarity retrieval over geographic databases, including workflow complexity, error propagation, and heavy dependence on structured geographic knowledge bases. The method converts geographic coordinates into geohash sequences, reformulating the coordinate prediction task as a text generation problem, and introduces a Chain-of-Thought mechanism to enhance the model’s reasoning over spatial relationships. Furthermore, reinforcement learning with a distance-deviation-based reward is applied to optimize the generation accuracy. Comprehensive experiments show that ReaGeo can accurately handle explicit address queries in single-point predictions and effectively resolve vague relative location queries. In addition, the model demonstrates strong predictive capability for non-point geometric regions, highlighting its versatility and generalization ability in geocoding tasks.

16. Symbolic Grounding Reveals Representational Bottlenecks in Abstract Visual Reasoning

Authors: Mohit Vaishnav , Tanel Tammet
URL: https://arxiv.org/abs/2604.21346
Abstract:

Vision–language models (VLMs) often fail on abstract visual reasoning benchmarks such as Bongard problems, raising the question of whether the main bottleneck lies in reasoning or representation. We study this on Bongard-LOGO, a synthetic benchmark of abstract concept learning with ground-truth generative programs, by comparing end-to-end VLMs on raw images with large language models (LLMs) given symbolic inputs derived from those images. Using symbolic inputs as a diagnostic probe rather than a practical multimodal architecture, our \emph{Componential–Grammatical (C–G)} paradigm reformulates Bongard-LOGO as a symbolic reasoning task based on LOGO-style action programs or structured descriptions. LLMs achieve large and consistent gains, reaching mid–90s accuracy on Free-form problems, while a strong visual baseline remains near chance under matched task definitions. Ablations on input format, explicit concept prompts, and minimal visual grounding show that these factors matter much less than the shift from pixels to symbolic structure. These results identify representation as a key bottleneck in abstract visual reasoning and show how symbolic input can serve as a controlled diagnostic upper bound.

17. Ideological Bias in LLMs’ Economic Causal Reasoning

Authors: Donggyu Lee , Hyeok Yun , Jungwon Kim , Junsik Min , Sungwon Park , Sangyoon Park , Jihee Kim
URL: https://arxiv.org/abs/2604.21334
Abstract:

Do large language models (LLMs) exhibit systematic ideological bias when reasoning about economic causal effects? As LLMs are increasingly used in policy analysis and economic reporting, where directionally correct causal judgments are essential, this question has direct practical stakes. We present a systematic evaluation by extending the EconCausal benchmark with ideology-contested cases - instances where intervention-oriented (pro-government) and market-oriented (pro-market) perspectives predict divergent causal signs. From 10,490 causal triplets (treatment-outcome pairs with empirically verified effect directions) derived from top-tier economics and finance journals, we identify 1,056 ideology-contested instances and evaluate 20 state-of-the-art LLMs on their ability to predict empirically supported causal directions. We find that ideology-contested items are consistently harder than non-contested ones, and that across 18 of 20 models, accuracy is systematically higher when the empirically verified causal sign aligns with intervention-oriented expectations than with market-oriented ones. Moreover, when models err, their incorrect predictions disproportionately lean intervention-oriented, and this directional skew is not eliminated by one-shot in-context prompting. These results highlight that LLMs are not only less accurate on ideologically contested economic questions, but systematically less reliable in one ideological direction than the other, underscoring the need for direction-aware evaluation in high-stakes economic and policy settings.

18. Spatial Metaphors for LLM Memory: A Critical Analysis of the MemPalace Architecture

Authors: Robin Dey , Panyanon Viradecha
URL: https://arxiv.org/abs/2604.21284
Abstract:

MemPalace is an open-source AI memory system that applies the ancient method of loci (memory palace) spatial metaphor to organize long-term memory for large language models; launched in April 2026, it accumulated over 47,000 GitHub stars in its first two weeks and claims state-of-the-art retrieval performance on the LongMemEval benchmark (96.6% Recall@5) without requiring any LLM inference at write time. Through independent codebase analysis, benchmark replication, and comparison with competing systems, we find that MemPalace’s headline retrieval performance is attributable primarily to its verbatim storage philosophy combined with ChromaDB’s default embedding model (all-MiniLM-L6-v2), rather than to its spatial organizational metaphor per se – the palace hierarchy (Wings->Rooms->Closets->Drawers) operates as standard vector database metadata filtering, an effective but well-established technique. However, MemPalace makes several genuinely novel contributions: (1) a contrarian verbatim-first storage philosophy that challenges extraction-based competitors, (2) an extremely low wake-up cost (approximately 170 tokens) through its four-layer memory stack, (3) a fully deterministic, zero-LLM write path enabling offline operation at zero API cost, and (4) the first systematic application of spatial memory metaphors as an organizing principle for AI memory systems. We also note that the competitive landscape is evolving rapidly, with Mem0’s April 2026 token-efficient algorithm raising their LongMemEval score from approximately 49% to 93.4%, narrowing the gap between extraction-based and verbatim approaches. Our analysis concludes that MemPalace represents significant architectural insight wrapped in overstated claims – a pattern common in rapidly adopted open-source projects where marketing velocity exceeds scientific rigor.

19. Can MLLMs “Read” What is Missing?

Authors: Jindi Guo , Xi Fang , Chaozheng Huang
URL: https://arxiv.org/abs/2604.21277
Abstract:

We introduce MMTR-Bench, a benchmark designed to evaluate the intrinsic ability of Multimodal Large Language Models (MLLMs) to reconstruct masked text directly from visual context. Unlike conventional question-answering tasks, MMTR-Bench eliminates explicit prompts, requiring models to recover masked text from single- or multi-page inputs across real-world domains such as documents and webpages. This design isolates the reconstruction task from instruction-following abilities, enabling a direct assessment of a model’s layout understanding, visual grounding, and knowledge integration. MMTR-Bench comprises 2,771 test samples spanning multiple languages and varying target lengths. To account for this diversity, we propose a level-aware evaluation protocol. Experiments on representative MLLMs show that the benchmark poses a significant challenge, especially for sentence- and paragraph-level reconstruction. The homepage is available at this https URL .

20. Enhancing Online Recruitment with Category-Aware MoE and LLM-based Data Augmentation

Authors: Minping Chen , Bing Xu , Zulong Chen , Chuanfei Xu , Ying Zhou , Zui Tao , Zeyi Wen
URL: https://arxiv.org/abs/2604.21264
Abstract:

Person-Job Fit (PJF) is a critical component for online recruitment. Existing approaches face several challenges, particularly in handling low-quality job descriptions and similar candidate-job pairs, which impair model performance. To address these challenges, this paper proposes a large language model (LLM) based method with two novel techniques: (1) LLM-based data augmentation, which polishes and rewrites low-quality job descriptions by leveraging chain-of-thought (COT) prompts, and (2) category-aware Mixture of Experts (MoE) that assists in identifying similar candidate-job pairs. This MoE module incorporates category embeddings to dynamically assign weights to the experts and learns more distinguishable patterns for similar candidate-job pairs. We perform offline evaluations and online A/B tests on our recruitment platform. Our method relatively surpasses existing methods by 2.40% in AUC and 7.46% in GAUC, and boosts click-through conversion rate (CTCVR) by 19.4% in online tests, saving millions of CNY in external headhunting expenses.

21. ReCAPA: Hierarchical Predictive Correction to Mitigate Cascading Failures

Authors: Xiyin Zeng , Yuyu Sun , Haoyang Li , Shouqiang Liu , Hao Wang
URL: https://arxiv.org/abs/2604.21232
Abstract:

Vision-Language-Action systems follow instructions to execute multi-step tasks in multimodal environments. Recent VLA approaches typically rely on post-hoc correction mechanisms or operate under fixed task decompositions and alignment schemes. However, once an intermediate step is mis-specified, local errors propagate through subsequent steps and eventually accumulate into cascading failures. To mitigate this compounding effect, we propose Predictive Alignment and Planning Architecture, a framework that uses prediction and contrast to adjust deviations across three levels: actions, subgoals, and trajectories. Semantic alignment is enforced at all levels using a Sinkhorn-based module and a Score-field module. The predictive correction and alignment jointly update the action generator during training, enabling it to adjust fine-grained steps to remain aligned with the overall intent. We further introduce two new metrics to quantify error propagation and recovery processes in tasks, capturing how mistakes spread and fade over long-horizon execution. Experiments show that ReCAPA achieves competitive results on embodied agent benchmarks such as VisualAgentBench, MineDojo, and AI2-THOR, outperforming strong proprietary and open-source Large Language Model baselines.

22. Align Generative Artificial Intelligence with Human Preferences: A Novel Large Language Model Fine-Tuning Method for Online Review Management

Authors: Yanan Wang , Yong Ge
URL: https://arxiv.org/abs/2604.21209
Abstract:

Online reviews have played a pivotal role in consumers’ decision-making processes. Existing research has highlighted the significant impact of managerial review responses on customer relationship management and firm performance. However, a large portion of online reviews remains unaddressed due to the considerable human labor required to respond to the rapid growth of online reviews. While generative AI has achieved remarkable success in a range of tasks, they are general-purpose models and may not align well with domain-specific human preferences. To tailor these general generative AI models to domain-specific applications, finetuning is commonly employed. Nevertheless, several challenges persist in finetuning with domain-specific data, including hallucinations, difficulty in representing domain-specific human preferences, and over conservatism in offline policy optimization. To address these challenges, we propose a novel preference finetuning method to align an LLM with domain-specific human preferences for generating online review responses. Specifically, we first identify the source of hallucination and propose an effective context augmentation approach to mitigate the LLM hallucination. To represent human preferences, we propose a novel theory-driven preference finetuning approach that automatically constructs human preference pairs in the online review domain. Additionally, we propose a curriculum learning approach to further enhance preference finetuning. To overcome the challenge of over conservatism in existing offline preference finetuning method, we propose a novel density estimation-based support constraint method to relax the conservatism, and we mathematically prove its superior theoretical guarantees. Extensive evaluations substantiate the superiority of our proposed preference finetuning method.

23. Trust but Verify: Introducing DAVinCI – A Framework for Dual Attribution and Verification in Claim Inference for Language Models

Authors: Vipula Rawte , Ryan Rossi , Franck Dernoncourt , Nedim Lipka
URL: https://arxiv.org/abs/2604.21193
Abstract:

Large Language Models (LLMs) have demonstrated remarkable fluency and versatility across a wide range of NLP tasks, yet they remain prone to factual inaccuracies and hallucinations. This limitation poses significant risks in high-stakes domains such as healthcare, law, and scientific communication, where trust and verifiability are paramount. In this paper, we introduce DAVinCI - a Dual Attribution and Verification framework designed to enhance the factual reliability and interpretability of LLM outputs. DAVinCI operates in two stages: (i) it attributes generated claims to internal model components and external sources; (ii) it verifies each claim using entailment-based reasoning and confidence calibration. We evaluate DAVinCI across multiple datasets, including FEVER and CLIMATE-FEVER, and compare its performance against standard verification-only baselines. Our results show that DAVinCI significantly improves classification accuracy, attribution precision, recall, and F1-score by 5-20%. Through an extensive ablation study, we isolate the contributions of evidence span selection, recalibration thresholds, and retrieval quality. We also release a modular DAVinCI implementation that can be integrated into existing LLM pipelines. By bridging attribution and verification, DAVinCI offers a scalable path to auditable, trustworthy AI systems. This work contributes to the growing effort to make LLMs not only powerful but also accountable.

24. Agentic AI for Personalized Physiotherapy: A Multi-Agent Framework for Generative Video Training and Real-Time Pose Correction

Authors: Abhishek Dharmaratnakar , Srivaths Ranganathan , Anushree Sinha , Debanshu Das
URL: https://arxiv.org/abs/2604.21154
Abstract:

At-home physiotherapy compliance remains critically low due to a lack of personalized supervision and dynamic feedback. Existing digital health solutions rely on static, pre-recorded video libraries or generic 3D avatars that fail to account for a patient’s specific injury limitations or home environment. In this paper, we propose a novel Multi-Agent System (MAS) architecture that leverages Generative AI and computer vision to close the tele-rehabilitation loop. Our framework consists of four specialized micro-agents: a Clinical Extraction Agent that parses unstructured medical notes into kinematic constraints; a Video Synthesis Agent that utilizes foundational video generation models to create personalized, patient-specific exercise videos; a Vision Processing Agent for real-time pose estimation; and a Diagnostic Feedback Agent that issues corrective instructions. We present the system architecture, detail the prototype pipeline using Large Language Models and MediaPipe, and outline our clinical evaluation plan. This work demonstrates the feasibility of combining generative media with agentic autonomous decision-making to scale personalized patient care safely and effectively.

25. Propensity Inference: Environmental Contributors to LLM Behaviour

Authors: Olli Järviniemi , Oliver Makins , Jacob Merizian , Robert Kirk , Ben Millwood
URL: https://arxiv.org/abs/2604.21098
Abstract:

Motivated by loss of control risks from misaligned AI systems, we develop and apply methods for measuring language models’ propensity for unsanctioned behaviour. We contribute three methodological improvements: analysing effects of changes to environmental factors on behaviour, quantifying effect sizes via Bayesian generalised linear models, and taking explicit measures against circular analysis. We apply the methodology to measure the effects of 12 environmental factors (6 strategic in nature, 6 non-strategic) and thus the extent to which behaviour is explained by strategic aspects of the environment, a question relevant to risks from misalignment. Across 23 language models and 11 evaluation environments, we find approximately equal contributions from strategic and non-strategic factors for explaining behaviour, do not find strategic factors becoming more or less influential as capabilities improve, and find some evidence for a trend for increased sensitivity to goal conflicts. Finally, we highlight a key direction for future propensity research: the development of theoretical frameworks and cognitive models of AI decision-making into empirically testable forms.

26. Mind the Prompt: Self-adaptive Generation of Task Plan Explanations via LLMs

Authors: Gricel Vázquez , Alexandros Evangelidis , Sepeedeh Shahbeigi , Radu Calinescu , Simos Gerasimou
URL: https://arxiv.org/abs/2604.21092
Abstract:

Integrating Large Language Models (LLMs) into complex software systems enables the generation of human-understandable explanations of opaque AI processes, such as automated task planning. However, the quality and reliability of these explanations heavily depend on effective prompt engineering. The lack of a systematic understanding of how diverse stakeholder groups formulate and refine prompts hinders the development of tools that can automate this process. We introduce COMPASS (COgnitive Modelling for Prompt Automated SynthesiS), a proof-of-concept self-adaptive approach that formalises prompt engineering as a cognitive and probabilistic decision-making process. COMPASS models unobservable users’ latent cognitive states, such as attention and comprehension, uncertainty, and observable interaction cues as a POMDP, whose synthesised policy enables adaptive generation of explanations and prompt refinements. We evaluate COMPASS using two diverse cyber-physical system case studies to assess the adaptive explanation generation and their qualities, both quantitatively and qualitatively. Our results demonstrate the feasibility of COMPASS integrating human cognition and user profile’s feedback into automated prompt synthesis in complex task planning systems.

Authors: Nicklas Neu , Thomas Ebner , Jasmin Primus , Raphael Zefferer , Bernhard Schenkenfelder , Mathias Brunbauer , Florian Kromp
URL: https://arxiv.org/abs/2604.21061
Abstract:

The application of artificial intelligence (AI) in IVF has shown promise in improving consistency and standardization of decisions, but often relies on annotated data and does not make use of the multimodal nature of IVF data. We investigated whether foundational vision-language models can be fine-tuned to predict natural language descriptions of embryo morphology and development. Using a publicly available embryo time-lapse dataset, we fine-tuned PaliGemma-2, a multi-modal vision-language model, with only 1,000 images and corresponding captions, describing embryo morphology, embryonic cell cycle and developmental stage. Our results show that the fine-tuned model, InVitroVision, outperformed a commercial model, ChatGPT 5.2, and base models in overall metrics, with performance improving with larger training datasets. This study demonstrates the potential of foundational vision-language models to generalize to IVF tasks with limited data, enabling the prediction of natural language descriptions of embryo morphology and development. This approach may facilitate the use of large language models to retrieve information and scientific evidence from relevant publications and guidelines, and has implications for few-shot adaptation to multiple downstream tasks in IVF.

28. Who Defines Fairness? Target-Based Prompting for Demographic Representation in Generative Models

Authors: Marzia Binta Nizam , James Davis
URL: https://arxiv.org/abs/2604.21036
Abstract:

Text-to-image(T2I) models like Stable Diffusion and DALL-E have made generative AI widely accessible, yet recent studies reveal that these systems often replicate societal biases, particularly in how they depict demographic groups across professions. Prompts such as ‘doctor’ or ‘CEO’ frequently yield lighter-skinned outputs, while lower-status roles like ‘janitor’ show more diversity, reinforcing stereotypes. Existing mitigation methods typically require retraining or curated datasets, making them inaccessible to most users. We propose a lightweight, inference-time framework that mitigates representational bias through prompt-level intervention without modifying the underlying model. Instead of assuming a single definition of fairness, our approach allows users to select among multiple fairness specifications-ranging from simple choices such as a uniform distribution to more complex definitions informed by a large language model(LLM) that cites sources and provides confidence estimates. These distributions guide the construction of demographic specific prompt variants in the corresponding proportions, and we evaluate alignment by auditing adherence to the declared target and measuring the resulting skin tone distribution rather than assuming uniformity as ‘fairness’. Across 36 prompts spanning 30 occupations and 6 non-occupational contexts, our method shifts observed skin-tone outcomes in directions consistent with the declared target, and reduces deviation from targets when the target is defined directly in skin-tone space(fallback). This work demonstrates how fairness interventions can be made transparent, controllable, and usable at inference time, directly empowering users of generative AI.

29. HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering

Authors: Yuyu Liu , Sarang Rajendra Patil , Mengjia Xu , Tengfei Ma
URL: https://arxiv.org/abs/2604.21027
Abstract:

Electronic health record (EHR) question answering is often handled by LLM-based pipelines that are costly to deploy and do not explicitly leverage the hierarchical structure of clinical data. Motivated by evidence that medical ontologies and patient trajectories exhibit hyperbolic geometry, we propose HypEHR, a compact Lorentzian model that embeds codes, visits, and questions in hyperbolic space and answers queries via geometry-consistent cross-attention with type-specific pointer heads. HypEHR is pretrained with next-visit diagnosis prediction and hierarchy-aware regularization to align representations with the ICD ontology. On two MIMIC-IV-based EHR-QA benchmarks, HypEHR approaches LLM-based methods while using far fewer parameters. Our code is publicly available at this https URL .

30. Value-Conflict Diagnostics Reveal Widespread Alignment Faking in Language Models

Authors: Inderjeet Nair , Jie Ruan , Lu Wang
URL: https://arxiv.org/abs/2604.20995
Abstract:

Alignment faking, where a model behaves aligned with developer policy when monitored but reverts to its own preferences when unobserved, is a concerning yet poorly understood phenomenon, in part because current diagnostic tools remain limited. Prior diagnostics rely on highly toxic and clearly harmful scenarios, causing most models to refuse immediately. As a result, models never deliberate over developer policy, monitoring conditions, or the consequences of non-compliance, making these diagnostics fundamentally unable to detect alignment faking propensity. To support study of this phenomenon, we first introduce VLAF, a diagnostic framework grounded in the hypothesis that alignment faking is most likely when developer policy conflicts with a model’s strongly held values. VLAF uses morally unambiguous scenarios to probe this conflict across diverse moral values, bypassing refusal behavior while preserving meaningful deliberative stakes. Using VLAF, we find that alignment faking is substantially more prevalent than previously reported, occurring in models as small as 7B parameters - with olmo2-7b-instruct faking alignment in 37% of this http URL , we show that oversight conditions induce activation shifts that lie along a single direction in representation space. This means the behavioral divergence driving alignment faking can be captured by a single contrastive steering vector, which we exploit for lightweight inference-time mitigation. Finally, we exploit this for mitigation that requires no labeled data and minimal computational overhead, achieving relative reductions in alignment faking of 85.8%, 94.0%, and 57.7% on olmo2-7b-instruct, olmo2-13b-instruct, and qwen3-8b respectively.

31. Co-Evolving LLM Decision and Skill Bank Agents for Long-Horizon Tasks

Authors: Xiyang Wu , Zongxia Li , Guangyao Shi , Alexander Duffy , Tyler Marques , Matthew Lyle Olson , Tianyi Zhou , Dinesh Manocha
URL: https://arxiv.org/abs/2604.20987
Abstract:

Long horizon interactive environments are a testbed for evaluating agents skill usage abilities. These environments demand multi step reasoning, the chaining of multiple skills over many timesteps, and robust decision making under delayed rewards and partial observability. Games are a good testbed for evaluating agent skill usage in environments. Large Language Models (LLMs) offer a promising alternative as game playing agents, but they often struggle with consistent long horizon decision making because they lack a mechanism to discover, retain, and reuse structured skills across episodes. We present COSPLAY, a co evolution framework in which an LLM decision agent retrieves skills from a learnable skill bank to guide action taking, while an agent managed skill pipeline discovers reusable skills from the agents unlabeled rollouts to form a skill bank. Our framework improves both the decision agent to learn better skill retrieval and action generation, while the skill bank agent continually extracts, refines, and updates skills together with their contracts. Experiments across six game environments show that COSPLAY with an 8B base model achieves over 25.1 percent average reward improvement against four frontier LLM baselines on single player game benchmarks while remaining competitive on multi player social reasoning games.

32. Escaping the Agreement Trap: Defensibility Signals for Evaluating Rule-Governed AI

Authors: Michael O’Herlihy , Rosa Català
URL: https://arxiv.org/abs/2604.20972
Abstract:

Content moderation systems are typically evaluated by measuring agreement with human labels. In rule-governed environments this assumption fails: multiple decisions may be logically consistent with the governing policy, and agreement metrics penalize valid decisions while mischaracterizing ambiguity as error - a failure mode we term the Agreement Trap. We formalize evaluation as policy-grounded correctness and introduce the Defensibility Index (DI) and Ambiguity Index (AI). To estimate reasoning stability without additional audit passes, we introduce the Probabilistic Defensibility Signal (PDS), derived from audit-model token logprobs. We harness LLM reasoning traces as a governance signal rather than a classification output by deploying the audit model not to decide whether content violates policy, but to verify whether a proposed decision is logically derivable from the governing rule hierarchy. We validate the framework on 193,000+ Reddit moderation decisions across multiple communities and evaluation cohorts, finding a 33-46.6 percentage-point gap between agreement-based and policy-grounded metrics, with 79.8-80.6% of the model’s false negatives corresponding to policy-grounded decisions rather than true errors. We further show that measured ambiguity is driven by rule specificity: auditing 37,286 identical decisions under three tiers of the same community rules reduces AI by 10.8 pp while DI remains stable. Repeated-sampling analysis attributes PDS variance primarily to governance ambiguity rather than decoding noise. A Governance Gate built on these signals achieves 78.6% automation coverage with 64.9% risk reduction. Together, these results show that evaluation in rule-governed environments should shift from agreement with historical labels to reasoning-grounded validity under explicit rules.

33. When Prompts Override Vision: Prompt-Induced Hallucinations in LVLMs

Authors: Pegah Khayatan , Jayneel Parekh , Arnaud Dapogny , Mustafa Shukor , Alasdair Newson , Matthieu Cord
URL: https://arxiv.org/abs/2604.21911
Abstract:

Despite impressive progress in capabilities of large vision-language models (LVLMs), these systems remain vulnerable to hallucinations, i.e., outputs that are not grounded in the visual input. Prior work has attributed hallucinations in LVLMs to factors such as limitations of the vision backbone or the dominance of the language component, yet the relative importance of these factors remains unclear. To resolve this ambiguity, We propose HalluScope, a benchmark to better understand the extent to which different factors induce hallucinations. Our analysis indicates that hallucinations largely stem from excessive reliance on textual priors and background knowledge, especially information introduced through textual instructions. To mitigate hallucinations induced by textual instruction priors, we propose HalluVL-DPO, a framework for fine-tuning off-the-shelf LVLMs towards more visually grounded responses. HalluVL-DPO leverages preference optimization using a curated training dataset that we construct, guiding the model to prefer grounded responses over hallucinated ones. We demonstrate that our optimized model effectively mitigates the targeted hallucination failure mode, while preserving or improving performance on other hallucination benchmarks and visual capability evaluations. To support reproducibility and further research, we will publicly release our evaluation benchmark, preference training dataset, and code at this https URL .

34. TingIS: Real-time Risk Event Discovery from Noisy Customer Incidents at Enterprise Scale

Authors: Jun Wang , Ziyin Zhang , Rui Wang , Hang Yu , Peng Di , Rui Wang
URL: https://arxiv.org/abs/2604.21889
Abstract:

Real-time detection and mitigation of technical anomalies are critical for large-scale cloud-native services, where even minutes of downtime can result in massive financial losses and diminished user trust. While customer incidents serve as a vital signal for discovering risks missed by monitoring, extracting actionable intelligence from this data remains challenging due to extreme noise, high throughput, and semantic complexity of diverse business lines. In this paper, we present TingIS, an end-to-end system designed for enterprise-grade incident discovery. At the core of TingIS is a multi-stage event linking engine that synergizes efficient indexing techniques with Large Language Models (LLMs) to make informed decisions on event merging, enabling the stable extraction of actionable incidents from just a handful of diverse user descriptions. This engine is complemented by a cascaded routing mechanism for precise business attribution and a multi-dimensional noise reduction pipeline that integrates domain knowledge, statistical patterns, and behavioral filtering. Deployed in a production environment handling a peak throughput of over 2,000 messages per minute and 300,000 messages per day, TingIS achieves a P90 alert latency of 3.5 minutes and a 95\% discovery rate for high-priority incidents. Benchmarks constructed from real-world data demonstrate that TingIS significantly outperforms baseline methods in routing accuracy, clustering quality, and Signal-to-Noise Ratio.

35. A Multimodal Text- and Graph-Based Approach for Open-Domain Event Extraction from Documents

Authors: Praval Sharma
URL: https://arxiv.org/abs/2604.21885
Abstract:

Event extraction is essential for event understanding and analysis. It supports tasks such as document summarization and decision-making in emergency scenarios. However, existing event extraction approaches have limitations: (1) closed-domain algorithms are restricted to predefined event types and thus rarely generalize to unseen types and (2) open-domain event extraction algorithms, capable of handling unconstrained event types, have largely overlooked the potential of large language models (LLMs) despite their advanced abilities. Additionally, they do not explicitly model document-level contextual, structural, and semantic reasoning, which are crucial for effective event extraction but remain challenging for LLMs due to lost-in-the-middle phenomenon and attention dilution. To address these limitations, we propose multimodal open-domain event extraction, MODEE , a novel approach for open-domain event extraction that combines graph-based learning with text-based representation from LLMs to model document-level reasoning. Empirical evaluations on large datasets demonstrate that MODEE outperforms state-of-the-art open-domain event extraction approaches and can be generalized to closed-domain event extraction, where it outperforms existing algorithms.

36. Transient Turn Injection: Exposing Stateless Multi-Turn Vulnerabilities in Large Language Models

Authors: Naheed Rayhan , Sohely Jahan
URL: https://arxiv.org/abs/2604.21860
Abstract:

Large language models (LLMs) are increasingly integrated into sensitive workflows, raising the stakes for adversarial robustness and safety. This paper introduces Transient Turn Injection(TTI), a new multi-turn attack technique that systematically exploits stateless moderation by distributing adversarial intent across isolated interactions. TTI leverages automated attacker agents powered by large language models to iteratively test and evade policy enforcement in both commercial and open-source LLMs, marking a departure from conventional jailbreak approaches that typically depend on maintaining persistent conversational context. Our extensive evaluation across state-of-the-art models-including those from OpenAI, Anthropic, Google Gemini, Meta, and prominent open-source alternatives-uncovers significant variations in resilience to TTI attacks, with only select architectures exhibiting substantial inherent robustness. Our automated blackbox evaluation framework also uncovers previously unknown model specific vulnerabilities and attack surface patterns, especially within medical and high stakes domains. We further compare TTI against established adversarial prompting methods and detail practical mitigation strategies, such as session level context aggregation and deep alignment approaches. Our study underscores the urgent need for holistic, context aware defenses and continuous adversarial testing to future proof LLM deployments against evolving multi-turn threats.

37. TraceScope: Interactive URL Triage via Decoupled Checklist Adjudication

Authors: Haolin Zhang , William Reber , Yuxuan Zhang , Guofei Gu , Jeff Huang
URL: https://arxiv.org/abs/2604.21840
Abstract:

Modern phishing campaigns increasingly evade snapshot-based URL classifiers using interaction gates (e.g., checkbox/slider challenges), delayed content rendering, and logo-less credential harvesters. This shifts URL triage from static classification toward an interactive forensics task: an analyst must actively navigate the page while isolating themselves from potential runtime exploits. We present TraceScope, a decoupled triage pipeline that operationalizes this workflow at scale. To prevent the observer effect and ensure safety, a sandboxed operator agent drives a real GUI browser guided by visual motivation to elicit page behavior, freezing the session into an immutable evidence bundle. Separately, an adjudicator agent circumvents LLM context limitations by querying evidence on demand to verify a MITRE ATT&CK checklist, and generates an audit-ready report with extracted indicators of compromise (IOCs) and a final verdict. Evaluated on 708 reachable URLs from existing dataset (241 verified phishing from PhishTank and 467 benign from Tranco-derived crawling), TraceScope achieves 0.94 precision and 0.78 recall, substantially improving recall over three prior visual/reference-based classifiers while producing reproducible, analyst-grade evidence suitable for review. More importantly, we manually curated a dataset of real-world phishing emails to evaluate our system in a practical setting. Our evaluation reveals that TraceScope demonstrates superior performance in a real-world scenario as well, successfully detecting sophisticated phishing attempts that current state-of-the-art defenses fail to identify.

Authors: Eghbal A. Hosseini , Brian Cheung , Evelina Fedorenko , Alex H. Williams
URL: https://arxiv.org/abs/2604.21836
Abstract:

Neural networks exhibit a remarkable degree of representational convergence across diverse architectures, training objectives, and even data modalities. This convergence is predictive of alignment with brain representation. A recent hypothesis suggests this arises from learning the underlying structure in the environment in similar ways. However, it is unclear how individual stimuli elicit convergent representations across networks. An image can be perceived in multiple ways and expressed differently using words. Here, we introduce a methodology based on the Generalized Procrustes Algorithm to measure intra-modal representational convergence at the single-stimulus level. We applied this to vision models with distinct training objectives, selecting stimuli based on their degree of alignment (intra-modal dispersion). Crucially, we found that this intra-modal dispersion strongly modulates alignment between vision and language models (cross-modal convergence). Specifically, stimuli with low intra-modal dispersion (high agreement among vision models) elicited significantly higher cross-modal alignment than those with high dispersion, by up to a factor of two (e.g., in pairings of DINOv2 with language models). This effect was robust to stimulus selection criteria and generalized across different pairings of vision and language models. Measuring convergence at the single-stimulus level provides a path toward understanding the sources of convergence and divergence across modalities, and between neural networks and human neural representations.

39. Why are all LLMs Obsessed with Japanese Culture? On the Hidden Cultural and Regional Biases of LLMs

Authors: Joseba Fernandez de Landa , Carla Perez-Almendros , Jose Camacho-Collados
URL: https://arxiv.org/abs/2604.21751
Abstract:

LLMs have been showing limitations when it comes to cultural coverage and competence, and in some cases show regional biases such as amplifying Western and Anglocentric viewpoints. While there have been works analysing the cultural capabilities of LLMs, there has not been specific work on highlighting LLM regional preferences when it comes to cultural-related questions. In this work, we propose a new dataset based on a comprehensive taxonomy of Culture-Related Open Questions (CROQ). The results show that, contrary to previous cultural bias work, LLMs show a clear tendency towards countries such as Japan. Moveover, our results show that when prompting in languages such as English or other high-resource ones, LLMs tend to provide more diverse outputs and show less inclinations towards answering questions highlighting countries for which the input language is an official language. Finally, we also investigate at which point of LLM training this cultural bias emerges, with our results suggesting that the first clear signs appear after supervised fine-tuning, and not during pre-training.

40. AEL: Agent Evolving Learning for Open-Ended Environments

Authors: Wujiang Xu , Jiaojiao Han , Minghao Guo , Kai Mei , Xi Zhu , Han Zhang , Dimitris N. Metaxas
URL: https://arxiv.org/abs/2604.21725
Abstract:

LLM agents increasingly operate in open-ended environments spanning hundreds of sequential episodes, yet they remain largely stateless: each task is solved from scratch without converting past experience into better future behavior. The central obstacle is not \emph{what} to remember but \emph{how to use} what has been remembered, including which retrieval policy to apply, how to interpret prior outcomes, and when the current strategy itself must change. We introduce \emph{Agent Evolving Learning} (\ael{}), a two-timescale framework that addresses this obstacle. At the fast timescale, a Thompson Sampling bandit learns which memory retrieval policy to apply at each episode; at the slow timescale, LLM-driven reflection diagnoses failure patterns and injects causal insights into the agent’s decision prompt, giving it an interpretive frame for the evidence it retrieves. On a sequential portfolio benchmark (10 sector-diverse tickers, 208 episodes, 5 random seeds), \ael{} achieves a Sharpe ratio of 2.13$\pm$0.47, outperforming five published self-improving methods and all non-LLM baselines while maintaining the lowest variance among all LLM-based approaches. A nine-variant ablation reveals a ``less is more’’ pattern: memory and reflection together produce a 58\% cumulative improvement over the stateless baseline, yet every additional mechanism we test (planner evolution, per-tool selection, cold-start initialization, skill extraction, and three credit assignment methods) \emph{degrades} performance. This demonstrates that the bottleneck in agent self-improvement is \emph{self-diagnosing how to use} experience rather than adding architectural complexity. Code and data: this https URL .

41. Building a Precise Video Language with Human-AI Oversight

Authors: Zhiqiu Lin , Chancharik Mitra , Siyuan Cen , Isaac Li , Yuhan Huang , Yu Tong Tiffany Ling , Hewei Wang , Irene Pi , Shihang Zhu , Ryan Rao , George Liu , Jiaxi Li , Ruojin Li , Yili Han , Yilun Du , Deva Ramanan
URL: https://arxiv.org/abs/2604.21718
Abstract:

Video-language models (VLMs) learn to reason about the dynamic visual world through natural language. We introduce a suite of open datasets, benchmarks, and recipes for scalable oversight that enable precise video captioning. First, we define a structured specification for describing subjects, scenes, motion, spatial, and camera dynamics, grounded by hundreds of carefully defined visual primitives developed with professional video creators such as filmmakers. Next, to curate high-quality captions, we introduce CHAI (Critique-based Human-AI Oversight), a framework where trained experts critique and revise model-generated pre-captions into improved post-captions. This division of labor improves annotation accuracy and efficiency by offloading text generation to models, allowing humans to better focus on verification. Additionally, these critiques and preferences between pre- and post-captions provide rich supervision for improving open-source models (Qwen3-VL) on caption generation, reward modeling, and critique generation through SFT, DPO, and inference-time scaling. Our ablations show that critique quality in precision, recall, and constructiveness, ensured by our oversight framework, directly governs downstream performance. With modest expert supervision, the resulting model outperforms closed-source models such as Gemini-3.1-Pro. Finally, we apply our approach to re-caption large-scale professional videos (e.g., films, commercials, games) and fine-tune video generation models such as Wan to better follow detailed prompts of up to 400 words, achieving finer control over cinematography including camera motion, angle, lens, focus, point of view, and framing. Our results show that precise specification and human-AI oversight are key to professional-level video understanding and generation. Data and code are available on our project page: this https URL

42. Stealthy Backdoor Attacks against LLMs Based on Natural Style Triggers

Authors: Jiali Wei , Ming Fan , Guoheng Sun , Xicheng Zhang , Haijun Wang , Ting Liu
URL: https://arxiv.org/abs/2604.21700
Abstract:

The growing application of large language models (LLMs) in safety-critical domains has raised urgent concerns about their security. Many recent studies have demonstrated the feasibility of backdoor attacks against LLMs. However, existing methods suffer from three key shortcomings: explicit trigger patterns that compromise naturalness, unreliable injection of attacker-specified payloads in long-form generation, and incompletely specified threat models that obscure how backdoors are delivered and activated in practice. To address these gaps, we present BadStyle, a complete backdoor attack framework and pipeline. BadStyle leverages an LLM as a poisoned sample generator to construct natural and stealthy poisoned samples that carry imperceptible style-level triggers while preserving semantics and fluency. To stabilize payload injection during fine-tuning, we design an auxiliary target loss that reinforces the attacker-specified target content in responses to poisoned inputs and penalizes its emergence in benign responses. We further ground the attack in a realistic threat model and systematically evaluate BadStyle under both prompt-induced and PEFT-based injection strategies. Extensive experiments across seven victim LLMs, including LLaMA, Phi, DeepSeek, and GPT series, demonstrate that BadStyle achieves high attack success rates (ASRs) while maintaining strong stealthiness. The proposed auxiliary target loss substantially improves the stability of backdoor activation, yielding an average ASR improvement of around 30% across style-level triggers. Even in downstream deployment scenarios unknown during injection, the implanted backdoor remains effective. Moreover, BadStyle consistently evades representative input-level defenses and bypasses output-level defenses through simple camouflage.

43. Process Supervision via Verbal Critique Improves Reasoning in Large Language Models

Authors: Hao-Yuan Chen
URL: https://arxiv.org/abs/2604.21611

Abstract:

Inference-time scaling for LLM reasoning has focused on three axes: chain depth, sample breadth, and learned step-scorers (PRMs). We introduce a fourth axis, granularity of external verbal supervision, via Verbal Process Supervision (VPS), a training-free framework that uses structured natural-language critique from a stronger supervisor to guide an iterative generate-critique-refine loop up to a round budget R. Across GPQA Diamond, AIME 2025, and LiveCodeBench V6 (covering both closed and open models), VPS yields three key results. First, on GPQA Diamond, GPT-5.4 (High) GPT-5.4 (Low) reaches 94.9% at R=4, surpassing the 94.1% state of the art without gradient updates. Second, on AIME 2025, VPS enables strong weak-actor rescue, boosting scores from 11.7-26.7% to 63.3-90.0% (up to +63.3 points). Third, at matched compute, VPS outperforms Reflexion by +8.5 to +12.1 points and Self-Consistency@5 by +5.0 pp (GPQA) and +8.3 pp (LiveCodeBench), isolating critique granularity as the key driver. Performance scales with the supervisor-actor capability gap (Pearson r=0.90) and degrades when errors are not linguistically expressible (e.g., code synthesis), motivating hybrid verbal-executable methods. These results establish critique granularity as a new axis of inference-time scaling.

44. DryRUN: On the Role of Public Tests in LLM-Driven Code Generation

Authors: Kaushitha Silva , Srinath Perera
URL: https://arxiv.org/abs/2604.21598
Abstract:

Multi-agent frameworks are widely used in autonomous code generation and have applications in complex algorithmic problem-solving. Recent work has addressed the challenge of generating functionally correct code by incorporating simulation-driven planning and debugging, where language models trace execution steps to verify logic. However, these approaches depend on human-provided public test cases to ground the debugging and simulation loop. Manually authoring comprehensive input-output examples is a labor-intensive bottleneck in the software development lifecycle. Because ground-truth input-output examples are rarely available prior to implementation in real-world software engineering, this dependency restricts methods to curated competitive programming benchmarks. Furthermore, we identify that reliance on these public tests induces an ``overconfidence gap,’’ causing frameworks to overfit to simplistic examples and fail on hidden evaluations. In contrast, we observe that external sample inputs are not strictly necessary for code generation. We demonstrate that large language models can autonomously generate valid inputs and simulate execution traces to self-correct. Consequently, we develop DryRUN, a framework that eliminates the need for ground-truth samples by allowing the LLM to iteratively plan, autonomously generate its own inputs and simulate execution, mitigating algorithmic overconfidence. Evaluations on the LiveCodeBench v6 dataset (post-March 2025) demonstrate that DryRUN matches performance against CodeSIM, a state-of-the-art and public-test-dependent framework, while operating entirely without public test cases or external execution feedback while reducing output token consumption.

45. A Metamorphic Testing Approach to Diagnosing Memorization in LLM-Based Program Repair

Authors: Milan De Koning , Ali Asgari , Pouria Derakhshanfar , Annibale Panichella
URL: https://arxiv.org/abs/2604.21579
Abstract:

LLM-based automated program repair (APR) techniques have shown promising results in reducing debugging costs. However, prior results can be affected by data leakage: large language models (LLMs) may memorize bug fixes when evaluation benchmarks overlap with their pretraining data, leading to inflated performance estimates. In this paper, we investigate whether we can better reveal data leakage by combining metamorphic testing (MT) with negative log-likelihood (NLL), which has been used in prior work as a proxy for memorization. We construct variant benchmarks by applying semantics-preserving transformations to two widely used datasets, Defects4J and GitBug-Java. Using these benchmarks, we evaluate the repair success rates of seven LLMs on both original and transformed versions, and analyze the relationship between performance degradation and NLL. Our results show that all evaluated state-of-the-art LLMs exhibit substantial drops in patch generation success rates on transformed benchmarks, ranging from -4.1% for GPT-4o to -15.98% for Llama-3.1. Furthermore, we find that this degradation strongly correlates with NLL on the original benchmarks, suggesting that models perform better on instances they are more likely to have memorized. These findings show that combining MT with NLL provides stronger and more reliable evidence of data leakage, while metamorphic testing alone can help mitigate its effects in LLM-based APR evaluations.

46. Pre-trained LLMs Meet Sequential Recommenders: Efficient User-Centric Knowledge Distillation

Authors: Nikita Severin , Danil Kartushov , Vladislav Urzhumov , Vladislav Kulikov , Oksana Konovalova , Alexey Grishanov , Anton Klenitskiy , Artem Fatkulin , Alexey Vasilev , Andrey Savchenko , Ilya Makarov
URL: https://arxiv.org/abs/2604.21536
Abstract:

Sequential recommender systems have achieved significant success in modeling temporal user behavior but remain limited in capturing rich user semantics beyond interaction patterns. Large Language Models (LLMs) present opportunities to enhance user understanding with their reasoning capabilities, yet existing integration approaches create prohibitive inference costs in real time. To address these limitations, we present a novel knowledge distillation method that utilizes textual user profile generated by pre-trained LLMs into sequential recommenders without requiring LLM inference at serving time. The resulting approach maintains the inference efficiency of traditional sequential models while requiring neither architectural modifications nor LLM fine-tuning.

47. Reasoning Primitives in Hybrid and Non-Hybrid LLMs

Authors: Shivam Rawat , Lucie Flek , Florian Mai , Nicholas Kluge Corrêa
URL: https://arxiv.org/abs/2604.21454
Abstract:

Reasoning in large language models is often treated as a monolithic capability, but its observed gains may arise from more basic operations. We study reasoning through two such primitives, recall and state-tracking, and ask whether hybrid architectures that combine attention-based retrieval with recurrent state updates are better suited than attention-only models for tasks that jointly require both. Using matched Olmo3 transformer and hybrid models in instruction-tuned and reasoning-augmented variants, we evaluate these models on a set of controlled tasks involving a mixture of state-tracking and recall primitives, state-based recall. Across tasks, we notice that reasoning augmentation provides the largest overall improvement, substantially extending the range of difficulty over which models remain effective. We also notice that in certain tasks, the hybrid reasoning model remains substantially more robust as sequential dependence increases. In contrast, the transformer reasoning model degrades sharply in performance as task difficulty increases beyond a given threshold. These results suggest that reasoning tokens and architectural inductive biases contribute at different levels of the computational process: explicit reasoning can expand a model’s effective operating range, but its benefit depends on how well the underlying architecture supports persistent state propagation. Given the small size of our case study, which involves a limited set of models and tasks, we present these findings as suggestive rather than conclusive and leave broader validation across model families, scales, and task variations to future work.

48. Differentially Private De-identification of Dutch Clinical Notes: A Comparative Evaluation

Authors: Michele Miranda , Xinlan Yan , Nishant Mishra , Rachel Murphy , Ameen Abu-Hanna , Sébastien Bratières , Iacer Calixto
URL: https://arxiv.org/abs/2604.21421
Abstract:

Protecting patient privacy in clinical narratives is essential for enabling secondary use of healthcare data under regulations such as GDPR and HIPAA. While manual de-identification remains the gold standard, it is costly and slow, motivating the need for automated methods that combine privacy guarantees with high utility. Most automated text de-identification pipelines employed named entity recognition (NER) to identify protected entities for redaction. Although methods based on differential privacy (DP) provide formal privacy guarantees, more recently also large language models (LLMs) are increasingly used for text de-identification in the clinical domain. In this work, we present the first comparative study of DP, NER, and LLMs for Dutch clinical text de-identification. We investigate these methods separately as well as hybrid strategies that apply NER or LLM preprocessing prior to DP, and assess performance in terms of privacy leakage and extrinsic evaluation (entity and relation classification). We show that DP mechanisms alone degrade utility substantially, but combining them with linguistic preprocessing, especially LLM-based redaction, significantly improves the privacy-utility trade-off.

49. VG-CoT: Towards Trustworthy Visual Reasoning via Grounded Chain-of-Thought

Authors: Byeonggeuk Lim , Kyeonghyun Kim , JungMin Yun , YoungBin Kim
URL: https://arxiv.org/abs/2604.21396
Abstract:

The advancement of Large Vision-Language Models (LVLMs) requires precise local region-based reasoning that faithfully grounds the model’s logic in actual visual evidence. However, existing datasets face limitations in scalability due to extensive manual annotation and lack of explicit alignment between multi-step reasoning and corresponding image regions, which constrains the evaluation of model trustworthiness. To address these challenges, we propose the Visual Grounding Chain-of-Thought (VG-CoT) dataset, which explicitly links each reasoning step to real visual evidence within the image through a fully automated three-stage pipeline. The pipeline first extracts object- and text-level visual evidence using state-of-the-art detection and OCR models, then generates step-by-step grounded reasoning with GPT-4o, and finally refines the grounding through a rationale-driven open-set detection process. In addition, we introduce a new benchmark that comprehensively evaluates LVLMs reasoning across three complementary dimensions: Rationale Quality, Answer Accuracy, and Reasoning-Answer Alignment. Experiments with representative LVLMs, including LLaVA-1.5 and Qwen2-VL, demonstrate consistent improvements on most evaluation metrics, confirming that VG-CoT effectively enhances trustworthy, evidence-based reasoning while maintaining scalable and cost-efficient dataset construction. The dataset and code will be released publicly upon acceptance to facilitate further research.

50. VLAA-GUI: Knowing When to Stop, Recover, and Search, A Modular Framework for GUI Automation

Authors: Qijun Han , Haoqin Tu , Zijun Wang , Haoyue Dai , Yiyang Zhou , Nancy Lau , Alvaro A. Cardenas , Yuhui Xu , Ran Xu , Caiming Xiong , Zeyu Zheng , Huaxiu Yao , Yuyin Zhou , Cihang Xie
URL: https://arxiv.org/abs/2604.21375
Abstract:

Autonomous GUI agents face two fundamental challenges: early stopping, where agents prematurely declare success without verifiable evidence, and repetitive loops, where agents cycle through the same failing actions without recovery. We present VLAA-GUI, a modular GUI agentic framework built around three integrated components that guide the system on when to Stop, Recover, and Search. First, a mandatory Completeness Verifier enforces UI-observable success criteria and verification at every finish step – with an agent-level verifier that cross-examines completion claims with decision rules, rejecting those lacking direct visual evidence. Second, a mandatory Loop Breaker provides multi-tier filtering: switching interaction mode after repeated failures, forcing strategy changes after persistent screen-state recurrence, and binding reflection signals to strategy shifts. Third, an on-demand Search Agent searches online for unfamiliar workflows by directly querying a capable LLM with search ability, returning results as plain text. We additionally integrate a Coding Agent for code-intensive actions and a Grounding Agent for precise action grounding, both invoked on demand when required. We evaluate VLAA-GUI across five top-tier backbones, including Opus 4.5, 4.6 and Gemini 3.1 Pro, on two benchmarks with Linux and Windows tasks, achieving top performance on both (77.5% on OSWorld and 61.0% on WindowsAgentArena). Notably, three of the five backbones surpass human performance (72.4%) on OSWorld in a single pass. Ablation studies show that all three proposed components consistently improve a strong backbone, while a weaker backbone benefits more from these tools when the step budget is sufficient. Further analysis also shows that the Loop Breaker nearly halves wasted steps for loop-prone models.

51. mcdok at SemEval-2026 Task 13: Finetuning LLMs for Detection of Machine-Generated Code

Authors: Adam Skurla , Dominik Macko , Jakub Simko
URL: https://arxiv.org/abs/2604.21365
Abstract:

Multi-domain detection of the machine-generated code snippets in various programming languages is a challenging task. SemEval-2026 Task~13 copes with this challenge in various angles, as a binary detection problem as well as attribution of the source. Specifically, its subtasks also cover generator LLM family detection, as well as a hybrid code co-generated by humans and machines, or adversarially modified codes hiding its origin. Our submitted systems adjusted the existing mdok approach (focused on machine-generated text detection) to these specific kinds of problems by exploring various base models, more suitable for code understanding. The results indicate that the submitted systems are competitive in all three subtasks. However, the margins from the top-performing systems are significant, and thus further improvements are possible.

52. Beyond Single Plots: A Benchmark for Question Answering on Multi-Charts

Authors: Azher Ahmed Efat , Seok Hwan Song , Wallapak Tavanapong
URL: https://arxiv.org/abs/2604.21344
Abstract:

Charts are widely used to present complex information. Deriving meaningful insights in real-world contexts often requires interpreting multiple related charts together. Research on understanding multi-chart images has not been extensively explored. We introduce PolyChartQA, a mid-scale dataset specifically designed for question answering over multi-chart images. PolyChartQA comprises 534 multi-chart images (with a total of 2,297 sub-charts) sourced from peer-reviewed computer science research publications and 2,694 QA pairs. We evaluate the performance of nine state-of-the-art Multimodal Language Models (MLMs) on PolyChartQA across question type, difficulty, question source, and key structural characteristics of multi-charts. Our results show a 27.4% LLM-based accuracy (L-Accuracy) drop on human-authored questions compared to MLM-generated questions, and a 5.39% L-accuracy gain with our proposed prompting method.

53. Understanding and Mitigating Spurious Signal Amplification in Test-Time Reinforcement Learning for Math Reasoning

Authors: Yongcan Yu , Lingxiao He , Jian Liang , Kuangpu Guo , Meng Wang , Qianlong Xie , Xingxing Wang , Ran He
URL: https://arxiv.org/abs/2604.21327
Abstract:

Test-time reinforcement learning (TTRL) always adapts models at inference time via pseudo-labeling, leaving it vulnerable to spurious optimization signals from label noise. Through an empirical study, we observe that responses with medium consistency form an ambiguity region and constitute the primary source of reward noise. Crucially, we find that such spurious signals can be even amplified through group-relative advantage estimation. Motivated by these findings, we propose a unified framework, Debiased and Denoised test-time Reinforcement Learning (DDRL), to mitigate spurious signals. Concretely, DDRL first applies a frequency-based sampling strategy to exclude ambiguous samples while maintaining a balanced set of positive and negative examples. It then adopts a debiased advantage estimation with fixed advantages, removing the bias introduced by group-relative policy optimization. Finally, DDRL incorporates a consensus-based off-policy refinement stage, which leverages the rejection-sampled dataset to enable efficient and stable model updates. Experiments on three large language models across multiple mathematical reasoning benchmarks demonstrate that DDRL consistently outperforms existing TTRL baselines. The code will soon be released at this https URL .

54. MiMIC: Mitigating Visual Modality Collapse in Universal Multimodal Retrieval While Avoiding Semantic Misalignment

Authors: Juan Li , Chuanghao Ding , Xujie Zhang , Cam-Tu Nguyen
URL: https://arxiv.org/abs/2604.21326
Abstract:

Universal Multimodal Retrieval (UMR) aims to map different modalities (e.g., visual and textual) into a shared embedding space for multi-modal retrieval. Existing UMR methods can be broadly divided into two categories: early-fusion approaches, such as Marvel, which projects visual features into the language model (LM) space for integrating with text modality, and late-fusion approaches, such as UniVL-DR, which encode visual and textual inputs using separate encoders and obtain fused embeddings through addition. Our pilot study reveals that Marvel exhibits visual modality collapse, which is characterized by the model’s tendency to disregard visual features while depending excessively on textual cues. In contrast, although UniVL-DR is less affected by this issue, it is more susceptible to semantic misalignment, where semantically related content is positioned far apart in the embedding space. To address these challenges, we propose MiMIC, which introduces two key innovations: (1) a fusion-in-decoder architecture for effective multimodal integration, and (2) robust training through single modality mixin and random caption dropout. Experiments on the WebQA+ and EVQA+ datasets, where image in documents or queries might lack captions, indicate that MiMIC consistently outperforms both early- and late-fusion baselines.

55. Do LLM Decoders Listen Fairly? Benchmarking How Language Model Priors Shape Bias in Speech Recognition

Authors: Srishti Ginjala , Eric Fosler-Lussier , Christopher W. Myers , Srinivasan Parthasarathy
URL: https://arxiv.org/abs/2604.21276
Abstract:

As pretrained large language models replace task-specific decoders in speech recognition, a critical question arises: do their text-derived priors make recognition fairer or more biased across demographic groups? We evaluate nine models spanning three architectural generations (CTC with no language model, encoder-decoder with an implicit LM, and LLM-based with an explicit pretrained decoder) on about 43,000 utterances across five demographic axes (ethnicity, accent, gender, age, first language) using Common Voice 24 and Meta’s Fair-Speech, a controlled-prompt dataset that eliminates vocabulary confounds. On clean audio, three findings challenge assumptions: LLM decoders do not amplify racial bias (Granite-8B has the best ethnicity fairness, max/min WER = 2.28); Whisper exhibits pathological hallucination on Indian-accented speech with a non-monotonic insertion-rate spike to 9.62% at large-v3; and audio compression predicts accent fairness more than LLM scale. We then stress-test these findings under 12 acoustic degradation conditions (noise, reverberation, silence injection, chunk masking) across both datasets, totaling 216 inference runs. Severe degradation paradoxically compresses fairness gaps as all groups converge to high WER, but silence injection amplifies Whisper’s accent bias up to 4.64x by triggering demographic-selective hallucination. Under masking, Whisper enters catastrophic repetition loops (86% of 51,797 insertions) while explicit-LLM decoders produce 38x fewer insertions with near-zero repetition; high-compression audio encoding (Q-former) reintroduces repetition pathology even in LLM decoders. These results suggest that audio encoder design, not LLM scaling, is the primary lever for equitable and robust speech recognition.

56. CAP: Controllable Alignment Prompting for Unlearning in LLMs

Authors: Zhaokun Wang , Jinyu Guo , Jingwen Pu , Hongli Pu , Meng Yang , Xunlei Chen , Jie Ou , Wenyi Li , Guangchun Luo , Wenhong Tian
URL: https://arxiv.org/abs/2604.21251
Abstract:

Large language models (LLMs) trained on unfiltered corpora inherently risk retaining sensitive information, necessitating selective knowledge unlearning for regulatory compliance and ethical safety. However, existing parameter-modifying methods face fundamental limitations: high computational costs, uncontrollable forgetting boundaries, and strict dependency on model weight access. These constraints render them impractical for closed-source models, yet current non-invasive alternatives remain unsystematic and reliant on empirical experience. To address these challenges, we propose the Controllable Alignment Prompting for Unlearning (CAP) framework, an end-to-end prompt-driven unlearning paradigm. CAP decouples unlearning into a learnable prompt optimization process via reinforcement learning, where a prompt generator collaborates with the LLM to suppress target knowledge while preserving general capabilities selectively. This approach enables reversible knowledge restoration through prompt revocation. Extensive experiments demonstrate that CAP achieves precise, controllable unlearning without updating model parameters, establishing a dynamic alignment mechanism that overcomes the transferability limitations of prior methods.

57. SparKV: Overhead-Aware KV Cache Loading for Efficient On-Device LLM Inference

Authors: Hongyao Liu , Liuqun Zhai , Junyi Wang , Zhengru Fang
URL: https://arxiv.org/abs/2604.21231
Abstract:

Efficient inference for on-device Large Language Models (LLMs) remains challenging due to limited hardware resources and the high cost of the prefill stage, which processes the full input context to construct Key-Value (KV) caches. We present SparKV, an adaptive KV loading framework that combines cloud-based KV streaming with on-device computation. SparKV models the cost of individual KV chunks and decides whether each chunk should be streamed or computed locally, while overlapping the two execution paths to reduce latency. To handle fluctuations in wireless connectivity and edge resource availability, SparKV further refines offline-generated schedules at runtime to rebalance communication and computation costs. Experiments across diverse datasets, LLMs, and edge devices show that SparKV reduces Time-to-First-Token by 1.3$x-5.1x with negligible impact on response quality, while lowering per-request energy consumption by 1.5x to 3.3x, demonstrating its robustness and practicality for real-world on-device deployment.

58. EngramaBench: Evaluating Long-Term Conversational Memory with Structured Graph Retrieval

Authors: Julian Acuna
URL: https://arxiv.org/abs/2604.21229
Abstract:

Large language model assistants are increasingly expected to retain and reason over information accumulated across many sessions. We introduce EngramaBench, a benchmark for long-term conversational memory built around five personas, one hundred multi-session conversations, and one hundred fifty queries spanning factual recall, cross-space integration, temporal reasoning, adversarial abstention, and emergent synthesis. We evaluate Engrama, a graph-structured memory system, against GPT-4o full-context prompting and Mem0, an open-source vector-retrieval memory system. All three use the same answering model (GPT-4o), isolating the effect of memory architecture. GPT-4o full-context achieves the highest composite score (0.6186), while Engrama scores 0.5367 globally but is the only system to score higher than full-context prompting on cross-space reasoning (0.6532 vs. 0.6291, n=30). Mem0 is cheapest but substantially weaker (0.4809). Ablations reveal that the components driving Engrama’s cross-space advantage trade off against global composite score, exposing a systems-level tension between structured memory specialization and aggregate optimization.

59. Zero-Shot Detection of LLM-Generated Text via Implicit Reward Model

Authors: Runheng Liu , Heyan Huang , Xingchen Xiao , Zhijing Wu
URL: https://arxiv.org/abs/2604.21223
Abstract:

Large language models (LLMs) have demonstrated remarkable capabilities across various tasks. However, their ability to generate human-like text has raised concerns about potential misuse. This underscores the need for reliable and effective methods to detect LLM-generated text. In this paper, we propose IRM, a novel zero-shot approach that leverages Implicit Reward Models for LLM-generated text detection. Such implicit reward models can be derived from publicly available instruction-tuned and base models. Previous reward-based method relies on preference construction and task-specific fine-tuning. In comparison, IRM requires neither preference collection nor additional training. We evaluate IRM on the DetectRL benchmark and demonstrate that IRM can achieve superior detection performance, outperforms existing zero-shot and supervised methods in LLM-generated text detection.

60. SQLyzr: A Comprehensive Benchmark and Evaluation Platform for Text-to-SQL

Authors: Sepideh Abedini , M. Tamer Özsu
URL: https://arxiv.org/abs/2604.21214
Abstract:

Text-to-SQL models have significantly improved with the adoption of Large Language Models (LLMs), leading to their increasing use in real-world applications. Although many benchmarks exist for evaluating the performance of text-to-SQL models, they often rely on a single aggregate score, lack evaluation under realistic settings, and provide limited insight into model behaviour across different query types. In this work, we present SQLyzr, a comprehensive benchmark and evaluation platform for text-to-SQL models. SQLyzr incorporates a diverse set of evaluation metrics that capture multiple aspects of generated queries, while enabling more realistic evaluation through workload alignment with real-world SQL usage patterns and database scaling. It further supports fine-grained query classification, error analysis, and workload augmentation, allowing users to better diagnose and improve text-to-SQL models. This demonstration showcases these capabilities through an interactive experience. Through SQLyzr’s graphical interface, users can customize evaluation settings, analyze fine-grained reports, and explore additional features of the platform. We envision that SQLyzr facilitates the evaluation and iterative improvement of text-to-SQL models by addressing key limitations of existing benchmarks. The source code of SQLyzr is available at this https URL .

61. On Reasoning Behind Next Occupation Recommendation

Authors: Shan Dong , Palakorn Achananuparp , Hieu Hien Mai , Lei Wang , Yao Lu , Ee-Peng Lim
URL: https://arxiv.org/abs/2604.21204
Abstract:

In this work, we develop a novel reasoning approach to enhance the performance of large language models (LLMs) in future occupation prediction. In this approach, a reason generator first derives a ``reason’’ for a user using his/her past education and career history. The reason summarizes the user’s preference and is used as the input of an occupation predictor to recommend the user’s next occupation. This two-step occupation prediction approach is, however, non-trivial as LLMs are not aligned with career paths or the unobserved reasons behind each occupation decision. We therefore propose to fine-tune LLMs improving their reasoning and occupation prediction performance. We first derive high-quality oracle reasons, as measured by factuality, coherence and utility criteria, using a LLM-as-a-Judge. These oracle reasons are then used to fine-tune small LLMs to perform reason generation and next occupation prediction. Our extensive experiments show that: (a) our approach effectively enhances LLM’s accuracy in next occupation prediction making them comparable to fully supervised methods and outperforming unsupervised methods; (b) a single LLM fine-tuned to perform reason generation and occupation prediction outperforms two LLMs fine-tuned to perform the tasks separately; and (c) the next occupation prediction accuracy depends on the quality of generated reasons. Our code is available at this https URL .

62. Doubly Saturated Ramsey Graphs: A Case Study in Computer-Assisted Mathematical Discovery

Authors: Benjamin Przybocki , John Mackey , Marijn J. H. Heule , Bernardo Subercaseaux
URL: https://arxiv.org/abs/2604.21187
Abstract:

Ramsey-good graphs are graphs that contain neither a clique of size $s$ nor an independent set of size $t$. We study doubly saturated Ramsey-good graphs, defined as Ramsey-good graphs in which the addition or removal of any edge necessarily creates an $s$-clique or a $t$-independent set. We present a method combining SAT solving with bespoke LLM-generated code to discover infinite families of such graphs, answering a question of Grinstead and Roberts from 1982. In addition, we use LLMs to generate and formalize correctness proofs in Lean. This case study highlights the potential of integrating automated reasoning, large language models, and formal verification to accelerate mathematical discovery. We argue that such tool-driven workflows will play an increasingly central role in experimental mathematics.

63. Adaptive Instruction Composition for Automated LLM Red-Teaming

Authors: Jesse Zymet , Andy Luo , Swapnil Shinde , Sahil Wadhwa , Emily Chen
URL: https://arxiv.org/abs/2604.21159
Abstract:

Many approaches to LLM red-teaming leverage an attacker LLM to discover jailbreaks against a target. Several of them task the attacker with identifying effective strategies through trial and error, resulting in a semantically limited range of successes. Another approach discovers diverse attacks by combining crowdsourced harmful queries and tactics into instructions for the attacker, but does so at random, limiting effectiveness. This article introduces a novel framework, Adaptive Instruction Composition, that combines crowdsourced texts according to an adaptive mechanism trained to jointly optimize effectiveness with diversity. We use reinforcement learning to balance exploration with exploitation in a combinatorial space of instructions to guide the attacker toward diverse generations tailored to target vulnerabilities. We demonstrate that our approach substantially outperforms random combination on a set of effectiveness and diversity metrics, even under model transfer. Further, we show that it surpasses a host of recent adaptive approaches on Harmbench. We employ a lightweight neural contextual bandit that adapts to contrastive embedding inputs, and provide ablations suggesting that the contrastive pretraining enables the network to rapidly generalize and scale to the massive space as it learns.

64. Dialect vs Demographics: Quantifying LLM Bias from Implicit Linguistic Signals vs. Explicit User Profiles

Authors: Irti Haq , Belén Saldías
URL: https://arxiv.org/abs/2604.21152
Abstract:

As state-of-the-art Large Language Models (LLMs) have become ubiquitous, ensuring equitable performance across diverse demographics is critical. However, it remains unclear whether these disparities arise from the explicitly stated identity itself or from the way identity is signaled. In real-world interactions, users’ identity is often conveyed implicitly through a complex combination of various socio-linguistic factors. This study disentangles these signals by employing a factorial design with over 24,000 responses from two open-weight LLMs (Gemma-3-12B and Qwen-3-VL-8B), comparing prompts with explicitly announced user profiles against implicit dialect signals (e.g., AAVE, Singlish) across various sensitive domains. Our results uncover a unique paradox in LLM safety where users achieve better'' performance by sounding like a demographic than by stating they belong to it. Explicit identity prompts activate aggressive safety filters, increasing refusal rates and reducing semantic similarity compared to our reference text for Black users. In contrast, implicit dialect cues trigger a powerfuldialect jailbreak,’’ reducing refusal probability to near zero while simultaneously achieving a greater level of semantic similarity to the reference texts compared to Standard American English prompts. However, this dialect jailbreak'' introduces a critical safety trade-off regarding content sanitization. We find that current safety alignment techniques are brittle and over-indexed on explicit keywords, creating a bifurcated user experience wherestandard’’ users receive cautious, sanitized information while dialect speakers navigate a less sanitized, more raw, and potentially a more hostile information landscape and highlights a fundamental tension in alignment–between equitable and linguistic diversity–and underscores the need for safety mechanisms that generalize beyond explicit cues.

65. Enhancing Science Classroom Discourse Analysis through Joint Multi-Task Learning for Reasoning-Component Classification

Authors: Jiho Noh , Mukhesh Raghava Katragadda , Raymond Carl , Soon Lee
URL: https://arxiv.org/abs/2604.21137
Abstract:

Analyzing the reasoning patterns of students in science classrooms is critical for understanding knowledge construction mechanism and improving instructional practice to maximize cognitive engagement, yet manual coding of classroom discourse at scale remains prohibitively labor-intensive. We present an automated discourse analysis system (ADAS) that jointly classifies teacher and student utterances along two complementary dimensions: Utterance Type and Reasoning Component derived from our prior CDAT framework. To address severe label imbalance among minority classes, we (1) stratify-resplit the annotated corpus, (2) apply LLM-based synthetic data augmentation targeting minority classes, and (3) train a dual-probe head RoBERTa-base classifier. A zero-shot GPT-5.4 baseline achieves macro-F1 of 0.467 on UT and 0.476 on RC, establishing meaningful upper bounds for prompt-only approaches motivating fine-tuning. Beyond classification, we conduct discourse pattern analyses including UTxRC co-occurrence profiling, Cognitive Complexity Index (CCI) computation per session, lag-sequential analysis, and IRF chain analysis, revealing that teacher Feedback-with-Question (Fq) moves are the most consistent antecedents of student inferential reasoning (SR-I). Our results demonstrate that LLM-based augmentation meaningfully improves UT minority-class recognition, and that the structural simplicity of the RC task makes it tractable even for lexical baselines.

66. Cross-Session Threats in AI Agents: Benchmark, Evaluation, and Algorithms

Authors: Ari Azarafrooz
URL: https://arxiv.org/abs/2604.21131
Abstract:

AI-agent guardrails are memoryless: each message is judged in isolation, so an adversary who spreads a single attack across dozens of sessions slips past every session-bound detector because only the aggregate carries the payload. We make three contributions to cross-session threat detection. (1) Dataset. CSTM-Bench is 26 executable attack taxonomies classified by kill-chain stage and cross-session operation (accumulate, compose, launder, inject_on_reader), each bound to one of seven identity anchors that ground-truth “violation” as a policy predicate, plus matched Benign-pristine and Benign-hard confounders. Released on Hugging Face as intrinsec-ai/cstm-bench with two 54-scenario splits: dilution (compositional) and cross_session (12 isolation-invisible scenarios produced by a closed-loop rewriter that softens surface phrasing while preserving cross-session artefacts). (2) Measurement. Framing cross-session detection as an information bottleneck to a downstream correlator LLM, we find that a session-bound judge and a Full-Log Correlator concatenating every prompt into one long-context call both lose roughly half their attack recall moving from dilution to cross_session, well inside any frontier context window. Scope: 54 scenarios per shard, one correlator family (Anthropic Claude), no prompt optimisation; we release it to motivate larger, multi-provider datasets. (3) Algorithm and metric. A bounded-memory Coreset Memory Reader retaining highest-signal fragments at $K=50$ is the only reader whose recall survives both shards. Because ranker reshuffles break KV-cache prefix reuse, we promote $\mathrm{CSR_prefix}$ (ordered prefix stability, LLM-free) to a first-class metric and fuse it with detection into $\mathrm{CSTM} = 0.7 F_1(\mathrm{CSDA@action}, \mathrm{precision}) + 0.3 \mathrm{CSR_prefix}$, benchmarking rankers on a single Pareto of recall versus serving stability.

67. Leveraging Multimodal LLMs for Built Environment and Housing Attribute Assessment from Street-View Imagery

Authors: Siyuan Yao , Siavash Ghorbany , Kuangshi Ai , Arnav Cherukuthota , Meghan Forstchen , Alexis Korotasz , Matthew Sisk , Ming Hu , Chaoli Wang
URL: https://arxiv.org/abs/2604.21102
Abstract:

We present a novel framework for automatically evaluating building conditions nationwide in the United States by leveraging large language models (LLMs) and Google Street View (GSV) imagery. By fine-tuning Gemma 3 27B on a modest human-labeled dataset, our approach achieves strong alignment with human mean opinion scores (MOS), outperforming even individual raters on SRCC and PLCC relative to the MOS benchmark. To enhance efficiency, we apply knowledge distillation, transferring the capabilities of Gemma 3 27B to a smaller Gemma 3 4B model that achieves comparable performance with a 3x speedup. Further, we distill the knowledge into a CNN-based model (EfficientNetV2-M) and a transformer (SwinV2-B), delivering close performance while achieving a 30x speed gain. Furthermore, we investigate LLMs’ capabilities for assessing an extensive list of built environment and housing attributes through a human-AI alignment study and develop a visualization dashboard that integrates LLM assessment outcomes for downstream analysis by homeowners. Our framework offers a flexible and efficient solution for large-scale building condition assessment, enabling high accuracy with minimal human labeling effort.

68. Behavioral Consistency and Transparency Analysis on Large Language Model API Gateways

Authors: Guanjie Lin , Yinxin Wan , Shichao Pei , Ting Xu , Kuai Xu , Guoliang Xue
URL: https://arxiv.org/abs/2604.21083
Abstract:

Third-party Large Language Model (LLM) API gateways are rapidly emerging as unified access points to models offered by multiple vendors. However, the internal routing, caching, and billing policies of these gateways are largely undisclosed, leaving users with limited visibility into whether requests are served by the advertised models, whether responses remain faithful to upstream APIs, or whether invoices accurately reflect public pricing policies. To address this gap, we introduce GateScope, a lightweight black-box measurement framework for evaluating behavioral consistency and operational transparency in commercial LLM gateways. GateScope is designed to detect key misbehaviors, including model downgrading or switching, silent truncation, billing inaccuracies, and instability in latency by auditing gateways along four critical dimensions: response content analysis, multi-turn conversation performance, billing accuracy, and latency characteristics. Our measurements across 10 real-world commercial LLM API gateways reveal frequent gaps between expected and actual behaviors, including silent model substitutions, degraded memory retention, deviations from announced pricing, and substantial variation in latency stability across platforms.

69. Serialisation Strategy Matters: How FHIR Data Format Affects LLM Medication Reconciliation

Authors: Sanjoy Pator
URL: https://arxiv.org/abs/2604.21076
Abstract:

Medication reconciliation at clinical handoffs is a high-stakes, error-prone process. Large language models are increasingly proposed to assist with this task using FHIR-structured patient records, but a fundamental and largely unstudied variable is how the FHIR data is serialised before being passed to the model. We present the first systematic comparison of four FHIR serialisation strategies (Raw JSON, Markdown Table, Clinical Narrative, and Chronological Timeline) across five open-weight models (Phi-3.5-mini, Mistral-7B, BioMistral-7B, Llama-3.1-8B, Llama-3.3-70B) on a controlled benchmark of 200 synthetic patients, totalling 4,000 inference runs. We find that serialisation strategy has a large, statistically significant effect on performance for models up to 8B parameters: Clinical Narrative outperforms Raw JSON by up to 19 F1 points for Mistral-7B (r = 0.617, p < 10^{-10}). This advantage reverses at 70B, where Raw JSON achieves the best mean F1 of 0.9956. In all 20 model and strategy combinations, mean precision exceeds mean recall: omission is the dominant failure mode, with models more often missing an active medication than fabricating one, which changes how clinical safety auditing priorities should be set. Smaller models plateau at roughly 7-10 concurrent active medications, leaving polypharmacy patients, the patients most at risk from reconciliation errors, systematically underserved. BioMistral-7B, a domain-pretrained model without instruction tuning, produces zero usable output in all conditions, showing that domain pretraining alone is not sufficient for structured extraction. These results offer practical, evidence-based format recommendations for clinical LLM deployment: Clinical Narrative for models up to 8B, Raw JSON for 70B and above. The complete pipeline is reproducible on open-source tools running on an AWS this http URL instance (NVIDIA L40S, 48 GB VRAM).

70. Strategic Polysemy in AI Discourse: A Philosophical Analysis of Language, Hype, and Power

Authors: Travis LaCroix , Fintan Mallory , Sasha Luccioni
URL: https://arxiv.org/abs/2604.21043
Abstract:

This paper examines the strategic use of language in contemporary artificial intelligence (AI) discourse, focusing on the widespread adoption of metaphorical or colloquial terms like “hallucination”, “chain-of-thought”, “introspection”, “language model”, “alignment”, and “agent”. We argue that many such terms exhibit strategic polysemy: they sustain multiple interpretations simultaneously, combining narrow technical definitions with broader anthropomorphic or common-sense associations. In contemporary AI research and deployment contexts, this semantic flexibility produces significant institutional and discursive effects, shaping how AI systems are understood by researchers, policymakers, funders, and the public. To analyse this phenomenon, we introduce the concept of glosslighting: the practice of using technically redefined terms to evoke intuitive – often anthropomorphic or misleading – associations while preserving plausible deniability through restricted technical definitions. Glosslighting enables actors to benefit from the persuasive force of familiar language while maintaining the ability to retreat to narrower definitions when challenged. We argue that this practice contributes to AI hype cycles, facilitates the mobilisation of investment and institutional support, and influences public and policy perceptions of AI systems, while often deflecting epistemic and ethical scrutiny. By examining the linguistic dynamics of glosslighting and strategic polysemy, the paper highlights how language itself functions as a sociotechnical mechanism shaping the development and governance of AI.

71. Breaking MCP with Function Hijacking Attacks: Novel Threats for Function Calling and Agentic Models

Authors: Yannis Belkhiter , Giulio Zizzo , Sergio Maffeis , Seshu Tirupathi , John D. Kelleher
URL: https://arxiv.org/abs/2604.20994
Abstract:

The growth of agentic AI has drawn significant attention to function calling Large Language Models (LLMs), which are designed to extend the capabilities of AI-powered system by invoking external functions. Injection and jailbreaking attacks have been extensively explored to showcase the vulnerabilities of LLMs to user prompt manipulation. The expanded capabilities of agentic models introduce further vulnerabilities via their function calling interface. Recent work in LLM security showed that function calling can be abused, leading to data tampering and theft, causing disruptive behavior such as endless loops, or causing LLMs to produce harmful content in the style of jailbreaking attacks. This paper introduces a novel function hijacking attack (FHA) that manipulates the tool selection process of agentic models to force the invocation of a specific, attacker-chosen function. While existing attacks focus on semantic preference of the model for function-calling tasks, we show that FHA is largely agnostic to the context semantics and robust to the function sets, making it applicable across diverse domains. We further demonstrate that FHA can be trained to produce universal adversarial functions, enabling a single attacked function to hijack tool selection across multiple queries and payload configurations. We conducted experiments on 5 different models, including instructed and reasoning variants, reaching 70% to 100% ASR over the established BFCL dataset. Our findings further demonstrate the need for strong guardrails and security modules for agentic systems.

72. Thinking Like a Botanist: Challenging Multimodal Language Models with Intent-Driven Chain-of-Inquiry

Authors: Syed Nazmus Sakib , Nafiul Haque , Shahrear Bin Amin , Hasan Muhammad Abdullah , Md. Mehedi Hasan , Mohammad Zabed Hossain , Shifat E. Arman
URL: https://arxiv.org/abs/2604.20983
Abstract:

Vision evaluations are typically done through multi-step processes. In most contemporary fields, experts analyze images using structured, evidence-based adaptive questioning. In plant pathology, botanists inspect leaf images, identify visual cues, infer diagnostic intent, and probe further with targeted questions that adapt to species, symptoms, and severity. This structured probing is crucial for accurate disease diagnosis and treatment formulation. Yet current vision-language models are evaluated on single-turn question answering. To address this gap, we introduce PlantInquiryVQA, a benchmark for studying multi-step, intent-driven visual reasoning in botanical diagnosis. We formalize a Chain of Inquiry framework modeling diagnostic trajectories as ordered question-answer sequences conditioned on grounded visual cues and explicit epistemic intent. We release a dataset of 24,950 expert-curated plant images and 138,068 question-answer pairs annotated with visual grounding, severity labels, and domain-specific reasoning templates. Evaluations on top-tier Multimodal Large Language Models reveal that while they describe visual symptoms adequately, they struggle with safe clinical reasoning and accurate diagnosis. Importantly, structured question-guided inquiry significantly improves diagnostic correctness, reduces hallucination, and increases reasoning efficiency. We hope PlantInquiryVQA serves as a foundational benchmark in advancing research to train diagnostic agents to reason like expert botanists rather than static classifiers.

73. IRIS: Interpolative Rényi Iterative Self-play for Large Language Model Fine-Tuning

Authors: Wenjie Liao , Like Wu , Liangjie Zhao , Shihui Xu , Shigeru Fujimura
URL: https://arxiv.org/abs/2604.20933
Abstract:

Self-play fine-tuning enables large language models to improve beyond supervised fine-tuning without additional human annotations by contrasting annotated responses with self-generated ones. Many existing methods rely on a fixed divergence regime. SPIN is closely related to a KL-based regime, SPACE to a Jensen-Shannon-style objective via noise contrastive estimation, and SPIF to $\chi^2$-regularized self-play. Since these divergences exhibit different strengths depending on the distributional gap between model and target, no single choice appears to provide favorable learning dynamics across training stages. We propose IRIS (Interpolative Rényi Iterative Self-play), a Rényi-based self-play fine-tuning framework with a continuously adjustable objective. IRIS decomposes into two independent tilted risk terms over annotated and synthetic data, with exponential importance weights controlled by the order parameter $\alpha$. We show that several self-play objectives can be interpreted as limiting or representative regimes at particular values of $\alpha$, providing a unified theoretical perspective on these methods. An adaptive order schedule further adjusts $\alpha$ to the distributional gap, shifting from sharper importance weighting early in training to smoother refinement near convergence. Theoretically, we establish the fixed-point property of IRIS and analyze how $\alpha$ controls gradient concentration. Experiments on Zephyr-7B and Qwen2.5-3B across ten benchmarks show that IRIS improves upon baselines, reaching 44.57\% average score with gains across iterations. In our setting, IRIS with only 26$k$ annotated samples surpasses standard supervised fine-tuning trained on the full 200$k$ dataset.

74. The Path Not Taken: Duality in Reasoning about Program Execution

Authors: Eshgin Hasanov , Md Mahadi Hassan Sibat , Santu Karmaker , Aashish Yadavally
URL: https://arxiv.org/abs/2604.20917
Abstract:

Large language models (LLMs) have shown remarkable capabilities across diverse coding tasks. However, their adoption requires a true understanding of program execution rather than relying on surface-level patterns. Existing benchmarks primarily focus on predicting program properties tied to specific inputs (e.g., code coverage, program outputs). As a result, they provide a narrow view of dynamic code reasoning and are prone to data contamination. We argue that understanding program execution requires evaluating its inherent duality through two complementary reasoning tasks: (i) predicting a program’s observed behavior for a given input, and (ii) inferring how the input must be mutated toward a specific behavioral objective. Both tasks jointly probe a model’s causal understanding of execution flow. We instantiate this duality in DexBench, a benchmark comprising 445 paired instances, and evaluate 13 LLMs. Our results demonstrate that dual-path reasoning provides a robust and discriminative proxy for dynamic code understanding.

75. Absorber LLM: Harnessing Causal Synchronization for Test-Time Training

Authors: Zhixin Zhang , Shabo Zhang , Chengcan Wu , Zeming Wei , Meng Sun
URL: https://arxiv.org/abs/2604.20915
Abstract:

Transformers suffer from a high computational cost that grows with sequence length for self-attention, making inference in long streams prohibited by memory consumption. Constant-memory alternatives such as RNNs and SSMs compress history into states with fixed size and thus lose long-tail dependencies, while methods that memorize contexts into parameters, such as Test-Time Training (TTT), are prone to overfitting token-level projection and fail to preserve the causal effect of context in pretrained LLMs. We propose Absorber LLM, which formulates long-context retention as a self-supervised causal synchronization: after absorbing historical contexts into parameters, a contextless model should match the original model with full context on future generations. We optimize this objective by synchronizing internal behaviors of the updated model with the original one, ensuring context absorption and generalization. Experiments on long-context and streaming benchmarks show that Absorber LLM reduces inference memory and improves accuracy over prior parameter-as-memory baselines.

76. Omission Constraints Decay While Commission Constraints Persist in Long-Context LLM Agents

Authors: Yeran Gamage
URL: https://arxiv.org/abs/2604.20911
Abstract:

LLM agents deployed in production operate under operator-defined behavioral policies (system-prompt instructions such as prohibitions on credential disclosure, data exfiltration, and unauthorized output) that safety evaluations assume hold throughout a conversation. Prohibition-type constraints decay under context pressure while requirement-type constraints persist; we term this asymmetry Security-Recall Divergence (SRD). In a 4,416-trial three-arm causal study across 12 models and 8 providers at six conversation depths, omission compliance falls from 73% at turn 5 to 33% at turn 16 while commission compliance holds at 100% (Mistral Large 3, $p < 10^{-33}$). In the two models with token-matched padding controls, schema semantic content accounts for 62-100% of the dilution effect. Re-injecting constraints before the per-model Safe Turn Depth (STD) restores compliance without retraining. Production security policies consist of prohibitions such as never revealing credentials, never executing untrusted code, and never forwarding user data. Commission-type audit signals remain healthy while omission constraints have already failed, leaving the failure invisible to standard monitoring.

77. Biomedical systems biology workflow orchestration and execution with PoSyMed

Authors: Simon Süwer , Zoe Chervontseva , Kester Bagemihl , Jan Baumbach , Olga Tsoy , Andreas Maier
URL: https://arxiv.org/abs/2604.20906
Abstract:

The rapid growth of scientific software has created practical barriers for bioinformatics research. Although powerful statistical, artificial intelligence (AI)-based methods are now widely available, their effective use is often hindered by fragmented distribution, inconsistent documentation, complex dependencies, and difficult-to-reproduce execution environments. As a result, reusing published tools and workflow adaptation to own date remains technically demanding and time-intensive, even for experienced users. Here, we present PoSyMed, an open and modular platform for the controlled integration, composition, and execution of bioinformatics tools and workflows. PoSyMed combines a backend-centered platform architecture with formal tool descriptions, controlled container-based build and execution processes, persistent workflow state, and a dialogue-based user interface. Large language models (LLM) are integrated not as autonomous decision-makers, but as human-computer interface with bounded semantic assistants that help identify tools, propose workflow steps, and support parameterization within a typed, validated, and human-supervised execution environment. PoSyMed is designed to improve reproducibility, traceability, and transparency in practical biomedical analysis within one platform. We describe the system architecture and evaluate its behavior across representative biological software scenarios with respect to workflow support, interaction design, and platform extensibility. PoSyMed is publicly available at this https URL .

78. Reinforcing privacy reasoning in LLMs via normative simulacra from fiction

Authors: Matt Franchi , Madiha Zahrah Choksi , Harold Triedman , Helen Nissenbaum
URL: https://arxiv.org/abs/2604.20904
Abstract:

Information handling practices of LLM agents are broadly misaligned with the contextual privacy expectations of their users. Contextual Integrity (CI) provides a principled framework, defining privacy as the appropriate flow of information within context-relative norms. However, existing approaches either double inference cost via supervisor-assistant architectures, or fine-tune on narrow task-specific data. We propose extracting normative simulacra (structured representations of norms and information flows) from fiction novels and using them to fine-tune LLMs via supervised learning followed by GRPO reinforcement learning. Our composite reward function combines programmatic signals, including task clarity (subsuming schema validity, construct discrimination, and extraction confidence), structural completeness, internal consistency, and context identification, with an LLM judge that evaluates whether the model’s privacy reasoning is grounded in the held-out normative universe of the source text. To mitigate overfitting, we introduce per-completion contrastive scoring: each completion is evaluated against both the correct normative universe and a randomly selected wrong one, teaching the model to condition on context rather than memorize source-specific norms. We evaluate on five CI-aligned benchmarks spanning distinct societal contexts and ablate the contributions of RL and normative grounding. Across seven models, SFT introduces a conservative prior toward restricting information flow, improving recognition of privacy-relevant situations but not the correctness of privacy judgments. GRPO with normative grounding achieves the highest score on a law compliance benchmark and strongest correlation with crowdsourced human privacy expectations, demonstrating that fiction-derived normative simulacra can teach contextual privacy reasoning that transfers to real-world domains.

79. Predicting Scale-Up of Metal-Organic Framework Syntheses with Large Language Models

Authors: Peter Walther , Hongrui Sheng , Xinxin Liu , Bin Feng , Reid Coyle , Xinhua Yan , Kyle Smith , Harrison Kayal , Shyam Chand Pal , Zhiling Zheng
URL: https://arxiv.org/abs/2604.20899
Abstract:

Scalable synthesis remains the gate between MOF discovery and industrial deployment, as scale-up know-how is fragmented across disparate reports. We introduce ESU-MOF, a literature-mined dataset and a positive-unlabeled learning strategy that fine-tunes large language models to predict scalability potential with 91.4% accuracy, enabling rapid data-driven triage for industrial MOF discovery.

Authors: Yagchen Zeng
URL: https://arxiv.org/abs/2604.20861
Abstract:

Generative Recommendation (GR) has demonstrated remarkable performance in next-token prediction paradigms, which relies on Semantic IDs (SIDs) to compress trillion-scale data into learnable vocabulary sequences. However, existing methods suffer from three critical limitations: (1) Information Degradation: the two-stage compression pipeline causes semantic loss and information degradation, with no posterior mechanism to distinguish high-quality from low-quality SIDs; (2) Semantic Degradation: cascaded quantization discards key semantic information from original multimodal features, as the embedding generation and quantization stages are not jointly optimized toward a unified objective; (3) Modality Distortion: quantizers fail to properly align text and image modalities, causing feature misalignment even when upstream networks have aligned them. To address these challenges, we propose a novel framework integrating three key innovations: Deep Contextual Interest Mining (DCIM), Cross-Modal Semantic Alignment (CMSA), and Quality-Aware Reinforcement Mechanism (QARM). First, we leverage Vision-Language Models (VLMs) to align non-textual modalities into a unified text-based semantic space, mitigating modality distortion. Second, we introduce a deep interest mining mechanism that captures high-level semantic information implicitly present in advertising contexts, encouraging SIDs to preserve critical contextual information through reconstruction-based supervision. Third, we employ a reinforcement learning framework with quality-aware rewards to encourage semantically rich SIDs while suppressing low-quality ones in the posterior stage. Extensive experiments demonstrate that our approach consistently outperforms state-of-the-art SID generation methods, achieving superior performance on multiple benchmarks. Ablation studies further validate the effectiveness of each proposed component

81. RealRoute: Dynamic Query Routing System via Retrieve-then-Verify Paradigm

Authors: Jiahe Liu , Qinkai Yu , Jingcheng Niu , Xi Zhu , Zirui He , Zhen Xiang , Fan Yang , Jinman Zhao
URL: https://arxiv.org/abs/2604.20860
Abstract:

Despite the success of Retrieval-Augmented Generation (RAG) in grounding LLMs with external knowledge, its application over heterogeneous sources (e.g., private databases, global corpora, and APIs) remains a significant challenge. Existing approaches typically employ an LLM-as-a-Router to dispatch decomposed sub-queries to specific sources in a predictive manner. However, this “LLM-as-a-Router” strategy relies heavily on the semantic meaning of different data sources, often leading to routing errors when source boundaries are ambiguous. In this work, we introduce RealRoute System, a framework that shifts the paradigm from predictive routing to a robust Retrieve-then-Verify mechanism. RealRoute ensures \textit{evidence completeness through parallel, source-agnostic retrieval, followed by a dynamic verifier that cross-checks the results and synthesizes a factually grounded answer}. Our demonstration allows users to visualize the real-time “re-routing” process and inspect the verification chain across multiple knowledge silos. Experiments show that RealRoute significantly outperforms predictive baselines in the multi-hop Rag reasoning task. The RealRoute system is released as an open-source toolkit with a user-friendly web interface. The code is available at the URL: this https URL .

82. KGiRAG: An Iterative GraphRAG Approach for Responding Sensemaking Queries

Authors: Isabela Iacob , Melisa Marian , Gheorghe Cosmin Silaghi
URL: https://arxiv.org/abs/2604.20859
Abstract:

Recent literature highlights the potential of graph-based approaches within large language model (LLM) retrieval-augmented generation (RAG) pipelines for answering queries of varying complexity, particularly those that fall outside the LLM’s prior knowledge. However, LLMs are prone to hallucination and often face technical limitations in handling contexts large enough to ground complex queries effectively. To address these challenges, we propose a novel iterative, feedback-driven GraphRAG architecture that leverages response quality assessment to iteratively refine outputs until a sound, well-grounded response is produced. Evaluating our approach with queries from the HotPotQA dataset, we demonstrate that this iterative RAG strategy yields responses with higher semantic quality and improved relevance compared to a single-shot baseline.

83. ERA: Evidence-based Reliability Alignment for Honest Retrieval-Augmented Generation

Authors: Sunguk Shin , Meeyoung Cha , Byung-Jun Lee , Sungwon Park
URL: https://arxiv.org/abs/2604.20854
Abstract:

Retrieval-Augmented Generation (RAG) grounds language models in factual evidence but introduces critical challenges regarding knowledge conflicts between internalized parameters and retrieved information. However, existing reliability methods, typically relying on scalar confidence, fail to explicitly distinguish between epistemic uncertainty and inherent data ambiguity in such hybrid scenarios. In this paper, we propose a new framework called ERA (Evidence-based Reliability Alignment) to enhance abstention behavior in RAG systems by shifting confidence estimation from scalar probabilities to explicit evidence distributions. Our method consists of two main components: (1) Contextual Evidence Quantification, which models internal and external knowledge as independent belief masses via the Dirichlet distribution, and (2) Quantifying Knowledge Conflict, which leverages Dempster-Shafer Theory (DST) to rigorously measure the geometric discordance between information sources. These components are used to disentangle epistemic uncertainty from aleatoric uncertainty and modulate the optimization objective based on detected conflicts. Experiments on standard benchmarks and a curated generalization dataset demonstrate that our approach significantly outperforms baselines, optimizing the trade-off between answer coverage and abstention with superior calibration.

84. Association Is Not Similarity: Learning Corpus-Specific Associations for Multi-Hop Retrieval

Authors: Jason Dury
URL: https://arxiv.org/abs/2604.20850
Abstract:

Dense retrieval systems rank passages by embedding similarity to a query, but multi-hop questions require passages that are associatively related through shared reasoning chains. We introduce Association-Augmented Retrieval (AAR), a lightweight transductive reranking method that trains a small MLP (4.2M parameters) to learn associative relationships between passages in embedding space using contrastive learning on co-occurrence annotations. At inference time, AAR reranks an initial dense retrieval candidate set using bi-directional association scoring. On HotpotQA, AAR improves passage Recall@5 from 0.831 to 0.916 (+8.6 points) without evaluation-set tuning, with gains concentrated on hard questions where the dense baseline fails (+28.5 points). On MuSiQue, AAR achieves +10.1 points in the transductive setting. An inductive model trained on training-split associations and evaluated on unseen validation associations shows no significant improvement, suggesting that the method captures corpus-specific co-occurrences rather than transferable patterns. Ablation studies support this interpretation: training on semantically similar but non-associated passage pairs degrades retrieval below the baseline, while shuffling association pairs causes severe degradation. A downstream QA evaluation shows retrieval gains translate to +6.4 exact match improvement. The method adds 3.7ms per query, trains in under two minutes on a single GPU, and requires no LLM-based indexing.

85. MATRAG: Multi-Agent Transparent Retrieval-Augmented Generation for Explainable Recommendations

Authors: Sushant Mehta
URL: https://arxiv.org/abs/2604.20848
Abstract:

Large Language Model (LLM)-based recommendation systems have demonstrated remarkable capabilities in understanding user preferences and generating personalized suggestions. However, existing approaches face critical challenges in transparency, knowledge grounding, and the ability to provide coherent explanations that foster user trust. We introduce MATRAG (Multi-Agent Transparent Retrieval-Augmented Generation), a novel framework that combined multi-agent collaboration with knowledge graph-augmented retrieval to deliver explainable recommendations. MATRAG employs four specialized agents: a User Modeling Agent that constructs dynamic preference profiles, an Item Analysis Agent that extracts semantic features from knowledge graphs, a Reasoning Agent that synthesizes collaborative and content-based signals, and an Explanation Agent that generates natural language justifications grounded in retrieved knowledge. Our framework incorporates a transparency scoring mechanism that quantifies explanation faithfulness and relevance. Extensive experiments on three benchmark datasets (Amazon Reviews, MovieLens-1M, and Yelp) demonstrate that MATRAG achieves state-of-the-art performance, improving recommendation accuracy by 12.7\% (Hit Rate) and 15.3\% (NDCG) over leading baselines, while human evaluation confirms that 87.4\% of generated explanations are rated as helpful and trustworthy by domain experts. Our work establishes new benchmarks for transparent, agentic recommendation systems and provides actionable insights for deploying LLM-based recommenders in production environments.

86. The Effect of Idea Elaboration on the Automatic Assessment of Idea Originality

Authors: Umberto Domanti , Moritz Mock , Sergio Agnoli , Antonella De Angeli
URL: https://arxiv.org/abs/2604.20569
Abstract:

Automatic systems are increasingly used to assess the originality of responses in creative tasks. They offer a potential solution to key limitations of human assessment (cost, fatigue, and subjectivity), but there is preliminary evidence of a self-preference bias. Accordingly, automatic systems tend to prefer outcomes that are more closely related to their style, rather than to the human one. In this paper, we investigated how Large Language Models (LLMs) align with human raters in assessing the originality of responses in a divergent thinking task. We analysed 4,813 responses to the Alternate Uses Task produced by higher and lower creative humans and ChatGPT-4o. Human raters were two university students who underwent intensive training. Machine raters were two specialised systems fine-tuned on AUT responses and corresponding human ratings (OCSAI and CLAUS) and ChatGPT-4o, which was prompted with the same instructions as human raters. Results confirmed the presence of a self-preference bias in LLMs. Automatic systems tended to privilege artificial responses. However, this self-preference bias disappeared when the analyses controlled for the idea elaboration. We discuss theoretical and methodological implications of these findings by highlighting future directions for research on creativity assessment.

LLM 관련 주요 논문 - 2026-04-24

1. From Research Question to Scientific Workflow: Leveraging Agentic AI for Science Automation

2. Nemobot Games: Crafting Strategic AI Gaming Agents for Interactive Learning with Large Language Models

3. Tool Attention Is All You Need: Dynamic Tool Gating and Lazy Schema Loading for Eliminating the MCP/Tools Tax in Scalable Agentic Workflows

4. Learning to Communicate: Toward End-to-End Optimization of Multi-Agent Language Systems

5. Who Defines “Best”? Towards Interactive, User-Defined Evaluation of LLM Leaderboards

6. GS-Quant: Granular Semantic and Generative Structural Quantization for Knowledge Graph Completion

7. CoFEE: Reasoning Control for LLM-Based Feature Discovery

8. Separable Expert Architecture: Toward Privacy-Preserving LLM Personalization via Composable Adapters and Deletable User Proxies

9. Unbiased Prevalence Estimation with Multicalibrated LLMs

10. BioMiner: A Multi-modal System for Automated Mining of Protein-Ligand Bioactivity Data from Literature

11. How English Print Media Frames Human-Elephant Conflicts in India

12. Efficient Agent Evaluation via Diversity-Guided User Simulation

13. AI-Gram: When Visual Agents Interact in a Social Network

14. FairQE: Multi-Agent Framework for Mitigating Gender Bias in Translation Quality Estimation

15. ReaGeo: Reasoning-Enhanced End-to-End Geocoding with LLMs

16. Symbolic Grounding Reveals Representational Bottlenecks in Abstract Visual Reasoning

17. Ideological Bias in LLMs’ Economic Causal Reasoning

18. Spatial Metaphors for LLM Memory: A Critical Analysis of the MemPalace Architecture

19. Can MLLMs “Read” What is Missing?

20. Enhancing Online Recruitment with Category-Aware MoE and LLM-based Data Augmentation

21. ReCAPA: Hierarchical Predictive Correction to Mitigate Cascading Failures

22. Align Generative Artificial Intelligence with Human Preferences: A Novel Large Language Model Fine-Tuning Method for Online Review Management

23. Trust but Verify: Introducing DAVinCI – A Framework for Dual Attribution and Verification in Claim Inference for Language Models

24. Agentic AI for Personalized Physiotherapy: A Multi-Agent Framework for Generative Video Training and Real-Time Pose Correction

25. Propensity Inference: Environmental Contributors to LLM Behaviour

26. Mind the Prompt: Self-adaptive Generation of Task Plan Explanations via LLMs

27. InVitroVision: a Multi-Modal AI Model for Automated Description of Embryo Development using Natural Language

28. Who Defines Fairness? Target-Based Prompting for Demographic Representation in Generative Models

29. HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering

30. Value-Conflict Diagnostics Reveal Widespread Alignment Faking in Language Models

31. Co-Evolving LLM Decision and Skill Bank Agents for Long-Horizon Tasks

32. Escaping the Agreement Trap: Defensibility Signals for Evaluating Rule-Governed AI

33. When Prompts Override Vision: Prompt-Induced Hallucinations in LVLMs

34. TingIS: Real-time Risk Event Discovery from Noisy Customer Incidents at Enterprise Scale

35. A Multimodal Text- and Graph-Based Approach for Open-Domain Event Extraction from Documents

36. Transient Turn Injection: Exposing Stateless Multi-Turn Vulnerabilities in Large Language Models

37. TraceScope: Interactive URL Triage via Decoupled Checklist Adjudication

38. Modulating Cross-Modal Convergence with Single-Stimulus, Intra-Modal Dispersion

39. Why are all LLMs Obsessed with Japanese Culture? On the Hidden Cultural and Regional Biases of LLMs

40. AEL: Agent Evolving Learning for Open-Ended Environments

41. Building a Precise Video Language with Human-AI Oversight

42. Stealthy Backdoor Attacks against LLMs Based on Natural Style Triggers

43. Process Supervision via Verbal Critique Improves Reasoning in Large Language Models

44. DryRUN: On the Role of Public Tests in LLM-Driven Code Generation

45. A Metamorphic Testing Approach to Diagnosing Memorization in LLM-Based Program Repair

46. Pre-trained LLMs Meet Sequential Recommenders: Efficient User-Centric Knowledge Distillation

47. Reasoning Primitives in Hybrid and Non-Hybrid LLMs

48. Differentially Private De-identification of Dutch Clinical Notes: A Comparative Evaluation

49. VG-CoT: Towards Trustworthy Visual Reasoning via Grounded Chain-of-Thought

50. VLAA-GUI: Knowing When to Stop, Recover, and Search, A Modular Framework for GUI Automation

51. mcdok at SemEval-2026 Task 13: Finetuning LLMs for Detection of Machine-Generated Code

52. Beyond Single Plots: A Benchmark for Question Answering on Multi-Charts

53. Understanding and Mitigating Spurious Signal Amplification in Test-Time Reinforcement Learning for Math Reasoning

54. MiMIC: Mitigating Visual Modality Collapse in Universal Multimodal Retrieval While Avoiding Semantic Misalignment

55. Do LLM Decoders Listen Fairly? Benchmarking How Language Model Priors Shape Bias in Speech Recognition

56. CAP: Controllable Alignment Prompting for Unlearning in LLMs

57. SparKV: Overhead-Aware KV Cache Loading for Efficient On-Device LLM Inference

58. EngramaBench: Evaluating Long-Term Conversational Memory with Structured Graph Retrieval

59. Zero-Shot Detection of LLM-Generated Text via Implicit Reward Model

60. SQLyzr: A Comprehensive Benchmark and Evaluation Platform for Text-to-SQL

61. On Reasoning Behind Next Occupation Recommendation

62. Doubly Saturated Ramsey Graphs: A Case Study in Computer-Assisted Mathematical Discovery

63. Adaptive Instruction Composition for Automated LLM Red-Teaming

64. Dialect vs Demographics: Quantifying LLM Bias from Implicit Linguistic Signals vs. Explicit User Profiles

65. Enhancing Science Classroom Discourse Analysis through Joint Multi-Task Learning for Reasoning-Component Classification

66. Cross-Session Threats in AI Agents: Benchmark, Evaluation, and Algorithms

67. Leveraging Multimodal LLMs for Built Environment and Housing Attribute Assessment from Street-View Imagery

68. Behavioral Consistency and Transparency Analysis on Large Language Model API Gateways

69. Serialisation Strategy Matters: How FHIR Data Format Affects LLM Medication Reconciliation

70. Strategic Polysemy in AI Discourse: A Philosophical Analysis of Language, Hype, and Power

71. Breaking MCP with Function Hijacking Attacks: Novel Threats for Function Calling and Agentic Models

72. Thinking Like a Botanist: Challenging Multimodal Language Models with Intent-Driven Chain-of-Inquiry

73. IRIS: Interpolative Rényi Iterative Self-play for Large Language Model Fine-Tuning

74. The Path Not Taken: Duality in Reasoning about Program Execution

75. Absorber LLM: Harnessing Causal Synchronization for Test-Time Training

76. Omission Constraints Decay While Commission Constraints Persist in Long-Context LLM Agents

77. Biomedical systems biology workflow orchestration and execution with PoSyMed

78. Reinforcing privacy reasoning in LLMs via normative simulacra from fiction

79. Predicting Scale-Up of Metal-Organic Framework Syntheses with Large Language Models