Papers

New on arXiv

Thu 30 Apr 2026

465 papers

Deterministic Replay for AI Agent SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
PlanFlip: Attacking Multi-Agent LLM Systems via Planning-Phase Prompt InjectionLLM agents and tool use → Agent threats, guardrails and governance +1
TokenScope: Token-Level Explainability and Interpretability for Code-Oriented Tasks in Large Language ModelsMechanistic interpretability → Knowledge and representations inside LLMs
Assert, don't describe: Linguistic features that shift LLM reasoning about animal welfareLLM reasoning → Reasoning benchmarks and evaluation
AONA: A Comprehensive Architecture and Workflow Design for Global Agentic CollaborationLLM agents and tool use → General LLM agent frameworks and benchmarks
Relation Reasoning with LLMs in Expensive OptimizationLLM reasoning → Test-time compute and efficient thinking
Data Deletion Can Help in Adaptive RLDeep reinforcement learning methods → Deep RL algorithms and representations
Agentic AI for Trip Planning Optimization ApplicationLLM agents and tool use → General LLM agent frameworks and benchmarks
When Do Diffusion Models learn to Generate Multiple Objects?Diffusion models → Diffusion foundations and sampling
Are You the A-hole? A Fair, Multi-Perspective Ethical Reasoning FrameworkLLM reasoning → Reasoning benchmarks and evaluation
Jailbroken Frontier Models Retain Their CapabilitiesLLM safety and security → Jailbreaking
Pessimism-Free Offline Learning in General-Sum Games via KL RegularizationOffline RL and imitation learning from data → Offline RL algorithms +1
Retrieval-Augmented Reasoning for Chartered AccountancyLLM reasoning → Reasoning for agents, planning and code +1
Rethinking Network Topologies for Cost-Effective Mixture-of-Experts LLM ServingMixture-of-experts → Efficient MoE training and inference systems
Lost in State Space: Probing Frozen Mamba RepresentationsState-space sequence models (S4, Mamba) → Core SSM architectures for language
Information-geometric adaptive sampling for graph diffusionDiffusion models → Flow matching, time series and science
Attention Is Where You AttackLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
TUR-DPO: Topology- and Uncertainty-Aware Direct Preference OptimizationLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
RSAT: Structured Attribution Makes Small Language Models Faithful Table ReasonersLLM reasoning → Test-time compute and efficient thinking
Diversity in Large Language Models under Supervised Fine-TuningLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Towards A Generative Protein Evolution Machine with DPLM-EvoDiffusion models → Flow matching, time series and science
Consistent Diffusion Language ModelsDiffusion models → Discrete and text diffusion
E2DT: Efficient and Effective Decision Transformer with Experience-Aware Sampling for Robotic ManipulationOffline RL and imitation learning from data → Real-world robot offline RL +1
Wasserstein Distributionally Robust Regret Optimization for Reinforcement Learning from Human FeedbackLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
Are Tools All We Need? Unveiling the Tool-Use Tax in LLM AgentsLLM agents and tool use → Tool learning and function calling
SPLICE: Latent Diffusion over JEPA Embeddings for Conformal Time-Series InpaintingDiffusion models → Flow matching, time series and science
Minimal, Local, Causal Explanations for Jailbreak Success in Large Language ModelsLLM safety and security → Jailbreaking
Cultural Benchmarking of LLMs in Standard and Dialectal Arabic DialoguesLLM reasoning → Reasoning benchmarks and evaluation
Alignment Contracts for Agentic Security SystemsLLM agents and tool use → Agent threats, guardrails and governance
World Model for Robot Learning: A Comprehensive SurveyWorld models → World models for navigation
Being-H0.7: A Latent World-Action Model from Egocentric VideosWorld models → Embodied and robot world models
AgentReputation: A Decentralized Agentic AI Reputation FrameworkLLM agents and tool use → Agent threats, guardrails and governance
Do Open-Loop Metrics Predict Closed-Loop Driving? A Cross-Benchmark Correlation Study of NAVSIM and Bench2DriveEnd-to-end autonomous driving → End-to-end system architecture
A Survey of Reasoning-Intensive Retrieval: Progress and ChallengesLLM reasoning → Knowledge- and tool-augmented reasoning
TADI: Tool-Augmented Drilling Intelligence via Agentic LLM Orchestration over Heterogeneous Wellsite DataLLM agents and tool use → Multimodal and embodied agents
Dynamic-TD3: A Novel Algorithm for UAV Path Planning with Dynamic Obstacle Trajectory PredictionDeep reinforcement learning methods → RL for robot navigation and UAV control
HERMES++: Toward a Unified Driving World Model for 3D Scene Understanding and GenerationWorld models → World models for autonomous driving +1
Visual Generation in the New Era: An Evolution from Atomic Mapping to Agentic World ModelingWorld models → Video-generation world models and simulators +2
Exploration Hacking: Can LLMs Learn to Resist RL Training?LLM reasoning → Reinforcement learning for reasoning +1
PhyCo: Learning Controllable Physical Priors for Generative MotionDiffusion models → Video generation
FlashRT: Towards Computationally and Memory Efficient Red-Teaming for Prompt Injection and Knowledge CorruptionRetrieval-augmented generation → Privacy and security of RAG +1
Global Optimality for Constrained Exploration via Penalty RegularizationDeep reinforcement learning methods → RL theory and sample complexity
Claw-Eval-Live: A Live Agent Benchmark for Evolving Real-World WorkflowsLLM agents and tool use → Software engineering coding agents
Crab: A Semantics-Aware Checkpoint/Restore Runtime for Agent SandboxesLLM agents and tool use → Software engineering coding agents
Latent Adversarial Detection: Adaptive Probing of LLM Activations for Multi-Turn Attack DetectionLLM safety and security → Jailbreaking
Beyond SFT-to-RL: Pre-alignment via Black-Box On-Policy Distillation for Multimodal RLMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +2
Beyond Gaussian Bottlenecks: Topologically Aligned Encoding of Vision-Transformer Feature SpacesWorld models → Video-generation world models and simulators
Do Sparse Autoencoders Capture Concept Manifolds?Mechanistic interpretability → Sparse autoencoder training and evaluation
GSDrive: Reinforcing Driving Policies by Multi-mode Future Trajectory Probing with 3D Gaussian Splatting EnvironmentEnd-to-end autonomous driving → Imitation-learning end-to-end planners
What Makes a Good Terminal-Agent Benchmark Task: A Guideline for Adversarial, Difficult, and Legible Evaluation DesignLLM agents and tool use → Agent skill learning and evolution
Intelligent Self-tuning Active EMI Filtering for Electrified Automotive Power Systems Using Reinforcement LearningDeep reinforcement learning methods → RL for energy, networks and operations
TopBench: A Benchmark for Implicit Predictive Reasoning in Tabular Question AnsweringLLM reasoning → Knowledge- and tool-augmented reasoning
Stable Behavior, Limited Variation: Persona Validity in LLM Agents for Urban Sentiment PerceptionLLM agents and tool use → Social simulation and multi-agent debate
Collaborative Agent Reasoning Engineering (CARE): A Three-Party Design Methodology for Systematically Engineering AI Agents with Subject Matter Experts, Developers, and Helper AgentsLLM reasoning → Reasoning for agents, planning and code +1
SpecVQA: A Benchmark for Spectral Understanding and Visual Question Answering in Scientific ImagesMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Echo-α: Large Agentic Multimodal Reasoning Model for Ultrasound InterpretationLLM reasoning → Clinical and biomedical applications +2
Learning from Disagreement: Clinician Overrides as Implicit Preference Signals for Clinical AI in Value-Based CareLLM alignment / RLHF / preference optimisation → Medical LLM alignment
Kernelized Advantage Estimation: From Nonparametric Statistics to LLM ReasoningLLM reasoning → Reinforcement learning for reasoning
A Pattern Language for Resilient Visual AgentsLLM agents and tool use → Multimodal and embodied agents
Latent-GRPO: Group Relative Policy Optimization for Latent ReasoningLLM reasoning → Reinforcement learning for reasoning
Exploring LLM Agent Designs and Interaction Modalities for Scientific VisualizationLLM agents and tool use → Multimodal and embodied agents
Dreaming Across Towns: Semantic Rollout and Town-Adversarial Regularization for Zero-Shot Held-Out-Town Fixed-Route Driving in CARLAEnd-to-end autonomous driving → Imitation-learning end-to-end planners
FineState-Bench: Benchmarking State-Conditioned Grounding for Fine-grained GUI State SettingMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
From LLM-Driven Trading Card Generation to Procedural Relatedness: A Pok'emon Case StudyDiffusion models → Image generation, text-to-image and editing
From Mirage to Grounding: Towards Reliable Multimodal Circuit-to-Verilog Code GenerationMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
LLMs as ASP Programmers: Self-Correction Enables Task-Agnostic Nonmonotonic ReasoningLLM reasoning → Reasoning benchmarks and evaluation
GUI Agents with Reinforcement Learning: Toward Digital InhabitantsLLM agents and tool use → GUI and mobile agents +1
The Effects of Visual Priming on Cooperative Behavior in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Diffusion-OAMP for Joint Image Compression and Wireless TransmissionDiffusion models → Diffusion for compression and wireless
Flying by Inference: Active Inference World Models for Adaptive UAV SwarmsWorld models → World models for navigation
MM-StanceDet: Retrieval-Augmented Multi-modal Multi-agent Stance DetectionRetrieval-augmented generation → Agentic RAG, agents and memory
Dynamic Cluster Data Sampling for Efficient and Long-Tail-Aware Vision-Language Pre-trainingMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Data-Driven Continuous-Time Linear Quadratic Regulator via Closed-Loop and Reinforcement Learning ParameterizationsDeep reinforcement learning methods → RL for games and combinatorial optimization
CoNewsReader: Supporting Comprehensive Understanding and Raising Critical Thoughts on Social Media News Through CommentsLLM reasoning → Reasoning benchmarks and evaluation
Graph World Models: Concepts, Taxonomy, and Future DirectionsWorld models → World models for language agents
Prediction-powered Inference by Mixture of ExpertsMixture-of-experts → MoE theory and gating functions
In-Context Prompting Obsoletes Agent Orchestration for Procedural TasksLLM agents and tool use → General LLM agent frameworks and benchmarks
Noise2Map: End-to-End Diffusion Model for Semantic Segmentation and Change DetectionDiffusion models → Inverse problems and image restoration
Building Persona-Based Agents On Demand: Tailoring Multi-Agent Workflows to User NeedsLLM agents and tool use → Social simulation and multi-agent debate
Modeling Clinical Concern Trajectories in Language Model AgentsLLM agents and tool use → Medical and scientific domain agents
TwinGate: Stateful Defense against Decompositional Jailbreaks in Untraceable Traffic via Asymmetric Contrastive LearningLLM safety and security → Jailbreaking
Rethinking Agentic Reinforcement Learning In Large Language ModelsLLM reasoning → Reasoning for agents, planning and code +1
NeocorRAG: Less Irrelevant Information, More Explicit Evidence, and More Effective Recall via Evidence ChainsRetrieval-augmented generation → RAG robustness and adaptive retrieval
Reasoning over Object Descriptions Improves Coreference Resolution in Task-Based Dialogue SystemsLLM reasoning → Reasoning benchmarks and evaluation
CastFlow: Learning Role-Specialized Agentic Workflows for Time Series ForecastingLLM agents and tool use → Financial trading and forecasting agents
ObjectGraph: From Document Injection to Knowledge Traversal -- A Native File Format for the Agentic EraLLM agents and tool use → Agent memory systems
Machine Unlearning for Class Removal through SISA-based Deep Neural Network ArchitecturesLLM safety and security → Machine unlearning
Motubrain: An Advanced World Action Model for Robot ControlWorld models → Embodied and robot world models +1
WindowsWorld: A Process-Centric Benchmark of Autonomous GUI Agents in Professional Cross-Application EnvironmentsLLM agents and tool use → GUI and mobile agents
Position-Aware Drafting for Inference Acceleration in LLM-Based Generative List-Wise RecommendationEfficient LLM inference and serving → Speculative decoding
Mind the Gap: Structure-Aware Consistency in Preference LearningLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
AgentEconomist: An End-to-end Agentic System Translating Economic Intuitions into Executable Computational ExperimentsLLM agents and tool use → Financial trading and forecasting agents
Iterative Multimodal Retrieval-Augmented Generation for Medical Question AnsweringRetrieval-augmented generation → Medical and biomedical RAG
Why Does Grounding Hurt Medical VQA? Benchmarking, Diagnosis, and Fine-Tuning of Vision-Language ModelsMultimodal LLMs / vision-language models → Medical vision-language models
Improving Calibration in Test-Time Prompt Tuning for Vision-Language Models via Data-Free Flatness-Aware Prompt PretrainingMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Knowledge Graph Representations for LLM-Based Policy Compliance ReasoningLLM reasoning → Reasoning for agents, planning and code
Contextual Agentic Memory is a Memo, Not True MemoryRetrieval-augmented generation → Agentic RAG, agents and memory +1
Bridging Values and Behavior: A Hierarchical Framework for Proactive Embodied AgentsLLM agents and tool use → Multimodal and embodied agents
When Agents Evolve, Institutions FollowLLM agents and tool use → Social simulation and multi-agent debate
PuzzleMark: Implicit Jigsaw Learning for Robust Code Dataset Watermarking in Neural Code Completion ModelsLLM safety and security → Watermarking generated content and models

All 465