Papers

New on arXiv

Fri 3 Jul 2026

328 papers

Margins, Not Windows: Training-Free Per-Step Lossy Speculative DecodingEfficient LLM inference and serving → Speculative decoding
SchemaRouter: Field-Aware Tool Routing for Efficient Heterogeneous Agentic RAGRetrieval-augmented generation → Agentic RAG, agents and memory +1
A Year in LLM Serving: Workload Evolution, Caching and Load-BalancingEfficient LLM inference and serving → Serving systems and request scheduling
HoosierHelp: Benchmarking LLM Agents for Social Service NavigationLLM agents and tool use → Social simulation and multi-agent debate
Reinforcement Learning for Heterogeneous Sensor Selection in Maritime SurveillanceDeep reinforcement learning methods → RL for robot navigation and UAV control
CacheSpec: Finding the Sweet Spot for Small Models in Large Language ModelsLLM agents and tool use → Software engineering coding agents
CrackedPDFs: A Controlled Benchmark for Hidden Prompt Injection in PDFsLLM safety and security → Prompt injection and agent security
From Trajectories to Prefixes: Reusing Teacher Trajectories via Replayed Prefixes and Online ContinuationLLM reasoning → Prompting and chain-of-thought reasoning +1
Replicating Belief, Not Bits: Epistemic State Replication for Agentic SystemsLLM agents and tool use → Agent memory systems
Scaffolding the Strategist: Architecture-Dependent Reasoning Interventions in Hotelling Spatial MarketsLLM reasoning → Reasoning for agents, planning and code
PORTS: Preference-Optimized Retrievers for Tool Selection with Large Language ModelsLLM agents and tool use → Tool learning and function calling +1
Modality Relevance is not Modality Utility: Post-hoc Selective Modality Escalation for Cost-Aware Multimodal RAGRetrieval-augmented generation → Multimodal RAG
Swarm-Driven Multi-Agent Reasoning for Smart City SecurityLLM agents and tool use → Agent threats, guardrails and governance
Reflected Schr"odinger Bridge MatchingDiffusion models → Diffusion foundations and sampling
RADIO1D: Elastic Representations for Condensed Vision ModelingMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
The Powerless Noise: How Experimental Settings Shape the Reported Power of NoiseRetrieval-augmented generation → RAG robustness and adaptive retrieval
IDEAL-Bench: Indoor Dataset and Evaluation suite for Analyzing 3D Layout reasoningLLM reasoning → Spatial and embodied reasoning +1
A Step Towards Robust Unsupervised Domain Adaptation via Fine-Tuning and Reinforcement LearningDeep reinforcement learning methods → Adversarial robustness and RL in finance
Token-Based Affordance Grounding with Large Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Responsibility Distribution Estimation in Ego-View Accident Videos with Multimodal Large Language ModelsMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection
Vision Non-Causal Trapezoidal Mamba: Eliminating Directional Scanning in Vision SSMs with Second-Order DynamicsState-space sequence models (S4, Mamba) → Vision Mamba backbones
When Geometry Aligns: Dihedral Hidden-State Transformations in UNet, ViT, and DiT ArchitecturesDiffusion models → Diffusion foundations and sampling
Teacher Supervision over Representation Equivalence ClassesKnowledge distillation → General knowledge distillation methods
Latent Clarity: Bridging World-Model Kinematics to Semantic Manifolds for Video Anomaly AnticipationWorld models → Video-generation world models and simulators
MentalThink: Shaping Thoughts in Mental SVG WorldLLM reasoning → Multimodal reasoning +1
Aligning Language Models with Selective PredictionLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
GameEngineBench: Evaluating Coding Agents on Real C++ Runtime EnvironmentsLLM agents and tool use → Software engineering coding agents
Perceptual Flow Matching for Few-Step Generative ModelingDiffusion models → Flow matching, time series and science
Anchored Self-Play for Code RepairSynthetic data and self-improvement for LLMs → Code and instruction data synthesis
Mixture-of-Gaussians-Guided Schedule Design for Brownian Bridge Diffusion ModelsDiffusion models → Inverse problems and image restoration
High-Precision Formation Control for Heterogeneous Multi-Robot Systems via Hierarchical Hybrid Physics-Informed Deep Reinforcement LearningDeep reinforcement learning methods → RL for robot navigation and UAV control
CAGE-1: Control, Assurance, and Governance Evaluation for Enterprise Agentic AILLM agents and tool use → Agent threats, guardrails and governance
Flex-Forcing: Towards a Unified Autoregressive and Bidirectional Video Diffusion ModelDiffusion models → Video generation
Reading Between the Dots: Decoding Hidden Computation across Filler TokensLLM reasoning → Test-time compute and efficient thinking
Learning to Generate Multiple Objects from Dense and Occluded LayoutsDiffusion models → Image generation, text-to-image and editing
MUTE: Return-Preserving Communication Unlearning for Efficient Multi-Agent CoordinationLLM safety and security → Machine unlearning
PhysMirror: Physics-Aware Mirror Object GenerationDiffusion models → 3D, point cloud and scene generation
WorldBagel: Uncovering the Power of Unified Multimodal Models for Vision-Language-Action-World ModelingWorld models → Embodied and robot world models +1
SentAttack: A Sentence-Level Black-Box Adversarial Attack Method for Dense Retrieval ModelsRetrieval-augmented generation → Privacy and security of RAG
Best-of-Better-N: Generating Pre-Aligned Responses with In-Context LearningLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
SkillOpt-Lite: Better and Faster Agent Self-evolution via One Line of VibeLLM agents and tool use → Agent skill learning and evolution
HiMe: Hierarchical Embodied Memory for Long-Horizon Vision-Language-Action ControlVision-language-action models → Reasoning and chain-of-thought in VLA
TRIAGE: Trustworthy Retrieval Instrumentation And Graph EvaluationRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
No Time Like the Present: Agentic Test-Time Training for LLM AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
Amortising Bayesian Experimental Design for Sequential Information Gathering in LLMsLLM reasoning → Reinforcement learning for reasoning
Securing Multi-Tool AI Agent Chains With Dynamic, Real-Time Compositional PoliciesLLM agents and tool use → Agent threats, guardrails and governance +1
Handwriting Trajectory Recovery with Diffusion ModelsDiffusion models → Image generation, text-to-image and editing
Efficient bias mitigation in T2I diffusion models using Concept GraphsDiffusion models → Image generation, text-to-image and editing +1
When Aggregate Alignment Misleads: Auditing Policy Repair Without Per-State Expert ActionsLLM agents and tool use → Agent threats, guardrails and governance
Brand-as-Memory: Vision-Language Models Encode Causal, Mechanistically Localizable Credibility Priors for News SourcesMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Pathways of Visual Information Flow in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
SPORK: Self-Speculative Forking to Accelerate Agentic LLM InferenceLLM agents and tool use → Software engineering coding agents
AtomicCommitBench: Can Coding Agents Reconstruct Commit Histories from Squashed Patches?LLM agents and tool use → Software engineering coding agents
From Judgments to Issues: Structured Extraction of Legal Reasoning with Citation-Hallucination ControlLLM reasoning → Reasoning for agents, planning and code
Is Agentic Code Review Helpful? Mining Developers' Feedback to CodeRabbit Reviews in the WildLLM agents and tool use → Software engineering coding agents
Piecewise Dynamic Diffusion Regularization for Reconstruction of Cardiac Cine MRIDiffusion models → Medical imaging and segmentation
A Decomposable Probe for Few-Step Diffusion Models: Prompt, Latent, and Score Selectivity across Backbone Families and Distillation ParadigmsDiffusion models → Diffusion foundations and sampling
Unbiased Alignment for Large Language Models with Noisy PreferencesLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
An Empirical Study of Downstream Adaptation for Agent SkillsLLM agents and tool use → Agent skill learning and evolution
TACG: Trajectory-Aware Commit Gating for Diffusion Language Model DecodingDiffusion models → Discrete and text diffusion
Agentic and Generative AI for Open-Source Intelligence and Cyber Investigations: Taxonomy, Evaluation, Challenges, and Future DirectionsLLM agents and tool use → Medical and scientific domain agents
Organizational Memory for Agentic Business Process ExecutionLLM agents and tool use → Software engineering coding agents
CONTRA: Red-Teaming Configurations of Personalizable AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
Reduced-Order Models: The Mother of World ModelsWorld models → Latent predictive world models
BVS: Bayesian Visual Search with Multimodal Large Language Model for Fine-grained PerceptionMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
AnchorVLA: Bridging Discrete Decisions and Continuous Trajectories for Vision-Language-Action PlanningVision-language-action models → VLA for autonomous driving
FairFlow: Demystifying and Mitigating Stereotype Bias in Text-to-Image Diffusion TransformersDiffusion models → Image generation, text-to-image and editing
CRRL: A Causality-Based Reinforcement Learning Framework for Autonomous System RecoveryDeep reinforcement learning methods → Deep RL algorithms and representations
Inverse Optimal Control for Linear Quadratic Problem with Poisson Jumps: Model-Free Inverse Reinforcement Learning ApproachesDeep reinforcement learning methods → RL for robot locomotion and manipulation
Entropy Regularization Improves Policy Robustness in Continuous-Time Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
KARMA: Knowledge graph-based Automated Reasoning Materialization and AlignmentLLM reasoning → Knowledge- and tool-augmented reasoning
APeB: Benchmarking Personalization Ability of Large Language Model AgentsLLM agents and tool use → Agent memory systems
DistillH-Mamba: A Hypergraph-Mamba-Based Knowledge Distillation Model for Efficient Impact Fall DetectionState-space sequence models (S4, Mamba) → Video, motion and generative Mamba models
Hope for the Best, Prepare for the Worst: Occlusion-Aware Contingency Planning for Autonomous VehiclesEnd-to-end autonomous driving → End-to-end system architecture
Exp2VLA: Enabling Vision-Language-Action for Drone Navigation from Expert DemonstrationsVision-language-action models → Aerial and drone VLA
Sample-Efficient Pareto Front Modeling for Energy-Aware Reinforcement Learning Using Bayesian OptimizationDeep reinforcement learning methods → RL for energy, networks and operations
Anticipatory Reinforcement Learning for Trajectory TrackingDeep reinforcement learning methods → RL for robot navigation and UAV control
ACPO: Asymmetric Credit Policy Optimization via Mode-Local Entropy SurrogateLLM reasoning → Reinforcement learning for reasoning +1
Integrating Physics-Informed Neural Networks for Safe Reinforcement Learning in a 1-DoF Helicopter SystemDeep reinforcement learning methods → Safe and constrained RL
Don't Wait to Reply: Towards Responsive yet Thoughtful Dialogue through Proactive ThinkingLLM reasoning → Audio-language model reasoning
SafeGuard: A Multi-Agent Perception-Reasoning Framework for Social-Risk AI-Generated Video DetectionLLM reasoning → Multimodal reasoning +1
Spectral Rewiring for Exploration, Purification, and Model MergingLLM reasoning → Reinforcement learning for reasoning
OmniFocus: Query-Guided Modality-Balanced Token Compression for Omni-Modal Large Language ModelsEfficient LLM inference and serving → Audio and omni-modal token compression
Compression, structure, and executor capability: a controlled real-cost decomposition of language-model agent skill optimisationLLM agents and tool use → Agent skill learning and evolution
HyperVAttention: Efficient Sparse Attention with Spatio-Temporal Clustering for Video DiffusionDiffusion models → Video generation
CONFLUX: A Latent Diffusion Model for 3D Chest-CT Synthesis with RL Post-TrainingDiffusion models → Medical imaging and segmentation
VISTA: Auditing Semantic Divergence in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
GuideMe: Multi-Domain Task Guidance and Intervention in Streaming VideoMultimodal LLMs / vision-language models → Video understanding with MLLMs
Reinforcement Learning for Evidence-Seeking Diagnostic Reasoning with Large Language ModelsLLM reasoning → Clinical and biomedical applications
Where Knowledge and Authority Sit Changes What an Agent Benchmark Can ResolveLLM agents and tool use → General LLM agent frameworks and benchmarks
Awakening Diffusion Transformers: Eliciting Stronger Generation and Understanding via Massive Activation ModulationDiffusion models → Diffusion foundations and sampling
Distill Where the Student Goes: Teacher-Regularized RL for English-Evidence Cross-Lingual RAGRetrieval-augmented generation → Generation and context integration +1
Individual Parameters in Weight-Sparse Transformers Appear InterpretableMechanistic interpretability → Circuits and mechanistic analysis
Parallelized Autoregressive Decoding for Omni-Modal Dense Video CaptioningEfficient LLM inference and serving → Fast autoregressive visual generation
Overloading Large Vision-Language Models for JailbreakingMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Incentivizing Vision Language Models to Search for Long Video Question AnsweringMultimodal LLMs / vision-language models → Video understanding with MLLMs +1
ReLo-IRR: Reflection-Guided LoRA Framework for Image Reflection RemovalDiffusion models → Inverse problems and image restoration
Serving Agentic Workflows with a Physical-Plan Compiler and Adaptive RuntimeLLM agents and tool use → Software engineering coding agents
A Sliding-Window-Based Reinforcement Learning for Dynamic Assembly Flow Shop Scheduling with Multi-Product DeliveryDeep reinforcement learning methods → RL for energy, networks and operations
CL-Anomaly: Layer-Adaptive Mixture-of-Experts with Multimodal Large Language Model for Continual Learning in Anomaly DetectionMultimodal LLMs / vision-language models → Chart understanding and anomaly detection +1

All 328