Papers

New on arXiv

Tue 28 Jul 2026

414 papers

Scene-Q: Confidence-Aware Coarse-to-Fine Querying of 3D Scenes with Selective VLM ReasoningLLM reasoning → Spatial and embodied reasoning
KDTwin: Task-Aware Knowledge Distillation for Lightweight Multi-Task Driving Scene SegmentationKnowledge distillation → Detection and dense prediction distillation
Alignment Is All You Need: Instruction-Free Training for General Audio-Language ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
Task- and Session-Level Model Routing: A Common-Interface Hybrid Evaluation of Four Open-Source Routers Across Four BenchmarksLLM agents and tool use → General LLM agent frameworks and benchmarks
Belayer: Efficient Fault Tolerance for LLM Agentic RL TrainingLLM agents and tool use → Reinforcement learning for LLM agents
Forecasting Side Effects of Activation SteeringMechanistic interpretability → Activation steering methods
SPOTting the Future: Lookahead Explanations for Deep Reinforcement LearningDeep reinforcement learning methods → Interpretable and logic-guided RL
Noise-Robust Conditional Flow Matching: Generating Clean Samples from Noisy DatasetsDiffusion models → Flow matching, time series and science
Meshy T2: Fast Native Mesh Generation with Flow MatchingDiffusion models → 3D, point cloud and scene generation
How Hard Does It Think? Analyzing Step-Aware Reasoning Energy in LLM Chain-of-Thought TrajectoriesLLM reasoning → Test-time compute and efficient thinking
Do Context Files Help Coding Agents? A Two-Agent Ablation Study on Real RepositoriesLLM agents and tool use → Software engineering coding agents
OwlPath: Lossless Knowledge Compression for LLM Bug RepairLLM agents and tool use → Software engineering coding agents
Divergence Decoding: Training-Free Capability FusionLLM reasoning → Test-time compute and efficient thinking
Incast-Free MoE Rate-Based SchedulingMixture-of-experts → Efficient MoE training and inference systems
RAGuard: A Layered Defense Framework for Retrieval-Augmented Generation Systems Against Data PoisoningRetrieval-augmented generation → Privacy and security of RAG +1
Learning Implicit Causal World Models from Multi-Agent DemonstrationsWorld models → World models for language agents
Seeing or Knowing? Visual Context Sensitivity in Multimodal Large Language ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
Route-Block Membership Selects Packed-AWQ Arithmetic: A Controlled Single-Fixture Mechanism StudyMixture-of-experts → Expert routing strategies
StealthBench: Measuring Operational Stealth in Autonomous Offensive-Security AgentsLLM agents and tool use → Agent threats, guardrails and governance
SARC-DQ: Runtime Data-Quality Gating for Agentic AI: Silent Evidence Defects, the Incompetence Shield, and Downstream-Only RemediationLLM agents and tool use → Agent threats, guardrails and governance
TraceCoder: Explainable and Auditable Code Generation with Position-Key Snippet VersioningLLM agents and tool use → Software engineering coding agents
MoSAIC: Aligned Intervention Supervision for Part-Local Motion Style TransferDiffusion models → Video generation
Denoising growth complexity: Data geometry and certified schedules for diffusion samplingDiffusion models → Diffusion foundations and sampling
Weak-to-Strong On-Policy DistillationKnowledge distillation → On-policy and self-distillation for LLMs
LumaGuide: Distribution Shaping for Training-Free HDR Generation in Diffusion ModelsDiffusion models → Inverse problems and image restoration
Multi-Agent Debate Strategies: Survey, Taxonomy, and ChallengesLLM agents and tool use → Social simulation and multi-agent debate
WildShadowRemover: In-the-Wild Video Shadow Removal via Detail-Preserving Video Diffusion ModelsDiffusion models → 3D, point cloud and scene generation
Rad-JEPA 3D: Radiology Joint-Embedding Predictive Model for 3D Computed TomographyState-space sequence models (S4, Mamba) → Mamba for medical image analysis
GoGoTB: Agentic RTL Verification with Specification-Grounded Coverage ClosureLLM agents and tool use → Software engineering coding agents
GuideSkill: Evolving Executable LLM Agent Skills for Guideline-Grounded Clinical ReasoningLLM reasoning → Clinical and biomedical applications +1
ClinLens: Towards Long-Horizon Coding Agents for Longitudinal Multimodal Clinical Data ScienceLLM agents and tool use → Medical and scientific domain agents
Even More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent SystemsLLM agents and tool use → Social simulation and multi-agent debate
Probing the Origins of Reasoning Performance: Representational Quality for Mathematical Problem-Solving in RL vs. SFT Fine-Tuned ModelsLLM reasoning → Reinforcement learning for reasoning
GPT-Red: Automated Red Teaming via Self-Play at ScaleSynthetic data and self-improvement for LLMs → Safety of self-evolving systems +1
A Reference-Free Score for Detecting Silent Reasoning Failures in Large Language ModelsLLM reasoning → Mathematical reasoning
Meta-Learned Reward Shaping for Reinforcement Learning from Human FeedbackLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
Pass the Baton: Trajectory-Relayed On-Policy DistillationKnowledge distillation → On-policy and self-distillation for LLMs
INTACT: Isomorphic Intent-to-Action Learning for Search-Free World ModelsWorld models → Embodied and robot world models
Spend Experts Where You Are Unsure: Confidence-Adaptive Routing for Mixture-of-Experts LoRAMixture-of-experts → Expert routing strategies
S2A2: Audio-Visual Imitation Learning for Manipulation Tasks Using Acoustic Spatial InformationOffline RL and imitation learning from data → Robot manipulation imitation learning
VetClaw: An Edge-Cloud Multimodal Agentic System for Veterinary Disease ScreeningLLM agents and tool use → Medical and scientific domain agents
Desktop-Delta Bench: Do Computer-Use Models Understand Desktop GUI Transitions?LLM agents and tool use → GUI and mobile agents
Reinformed Dreamer: An Asymmetric World Model Efficiently Trained through Latent GuidanceWorld models → Latent dynamics models and model-based RL +1
Wonder: Video World Model Done BetterWorld models → Video-generation world models and simulators
UniMem: Complementary Episodic-to-Parametric Memory for Boundary-Agnostic Task StreamsLLM agents and tool use → Agent memory systems
Pictura: Perspective-View Self-Play at Scale for DrivingEnd-to-end autonomous driving → Imitation-learning end-to-end planners
Parallel Decoding Distillation for Fast Image and Video GenerationDiffusion models → Video generation +1
RepoReasoner: Evaluating Repository-Level Code Reasoning Ability of Long-Context Language ModelsLLM reasoning → Reasoning for agents, planning and code
Beyond Zooming: Learning Multi-Tool Visual Reasoning for Ultra-High-Resolution Remote SensingLLM reasoning → Spatial and embodied reasoning +1
MemLens: A Value-Aware Memory Management System with Interactive Analytics for LLM-based AgentsLLM agents and tool use → Agent memory systems
Untangling Co-Drift: Proactive Multi-Intent Failure Prediction and Root-Cause Disambiguation for Self-Driving NetworksEnd-to-end autonomous driving → Reasoning VLA and VLM driving policies
Physics-Aware End-to-End Deep Reinforcement Learning for Quadcopter Control with Actuator DynamicsDeep reinforcement learning methods → RL for robot navigation and UAV control
Reinforcement Learning for Code OptimizationDeep reinforcement learning methods → RL training of language and agent models
LaP-Forensics: Latent-Pixel Consistency Guided Multimodal Reasoning for Deepfake DetectionLLM reasoning → Multimodal reasoning +1
Knowledge-Guided Multimodal Reasoning over Interacting Streams for Video-Level Ambivalence and Hesitancy RecognitionLLM reasoning → Audio-language model reasoning
Detecting Knowledge Inconsistencies Across Text, Tables, and Knowledge GraphsRetrieval-augmented generation → RAG robustness and adaptive retrieval
MODUS: Decoder-Only Any-to-Any Modeling of Diverse ModalitiesMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
A Cost-Effective Multimodal LLM Reasoning Framework for Question Answering over Irregular Clinical Time SeriesLLM reasoning → Clinical and biomedical applications +1
Evaluating Multi-Turn Multimodal Diagnostic Reasoning on Challenging Real-World Clinical CasesLLM reasoning → Clinical and biomedical applications
Can Deep Generative Models Reproduce Non-Stationary Gaussian Random Fields?Diffusion models → Flow matching, time series and science
Evaluating VLMs for Autonomous Agent-Driven Geometry Clipping Detection in Video Game QAMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
DC-WAM: Dynamic-Centric Visual Supervision and Reasoning for World-Action ModelsWorld models → Embodied and robot world models
Penelope: Localized Latent Recurrence for Efficient Structured ReasoningLLM reasoning → Test-time compute and efficient thinking
SAM3D-Guided Object-Centric Representation Alignment for Vision-Language-Action ModelsVision-language-action models → 3D and spatial-aware VLA
Minimizing Targeted Activations: Input-Only Suppression of Evaluation-Awareness Latents in Large Language ModelsMechanistic interpretability → Activation steering methods
Interactive Reward Agent: GUI Task Evaluation via Environment-State VerificationLLM agents and tool use → GUI and mobile agents
RecoReward: Recommender-Guided Multimodal Description Generation for RecommendationMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
TIGA: Trajectory-Injected Generative Attack against Black-box AIGC DetectorsDiffusion models → Image generation, text-to-image and editing
Distributing Security Controls Through Harness EngineeringLLM agents and tool use → Software engineering coding agents
RSIBench-Data: Benchmarking Data-Centric Research for Recursive Self-ImprovementSynthetic data and self-improvement for LLMs → Self-evolving LLM agents
Towards a Systems Foundation for Agentic Cloud ManagementLLM agents and tool use → Software engineering coding agents
Runtime Uncertainty Monitoring for LLM-Based Multi-Agent Systems Using Bayesian NetworksLLM agents and tool use → General LLM agent frameworks and benchmarks
HiSkill: Empowering LLM Agents with Hierarchical Skill GraphsLLM agents and tool use → Agent skill learning and evolution
AngelSpec: Towards Real-World High Performance Inference with Speculative DecodingEfficient LLM inference and serving → Speculative decoding
Rethinking Training Data for Generating Code Review CommentsLLM reasoning → Reasoning for agents, planning and code
Lowering the implementation barrier of neutral-atom quantum computing with agentic workflowsLLM agents and tool use → Medical and scientific domain agents
CHILL-Harness: Counterfactual Harness Learning for Efficient Reasoning in Long-Horizon AgentsLLM reasoning → Reasoning for agents, planning and code +1
SepPrune:A Separator-based Pruning Framework for Efficient Multimodal Large Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency
Speculate While You Reason: Teaching Agents to Predict Their Next Tool Call via Joint Agent-Speculator RLLLM reasoning → Reasoning for agents, planning and code +1
FLASH: Efficient Impact Fall Detection with Unified Hypergraph State-Space ModelState-space sequence models (S4, Mamba) → Video, motion and generative Mamba models
Tri-Manual Visuomotor Imitation Learning of Robot PoliciesOffline RL and imitation learning from data → Robot manipulation imitation learning
Shared Voxel-Map-Based Cooperative Indoor UAV Guidance with a Multi-Agent Soft Actor-Critic ControllerWorld models → World models for navigation
Tools Are Not Islands: Set-Level Tool Retrieval for LLM Agents via Query-Conditioned Hyperedge PredictionLLM agents and tool use → Tool learning and function calling
OmniDelta: Skill-Driven Budget Allocation for Token Compression in OmniLLMsEfficient LLM inference and serving → Audio and omni-modal token compression
CoRT: Counterfactual Replay for Token-Level Rubric-Guided Policy OptimizationLLM reasoning → Reinforcement learning for reasoning
An Empirical Study of Model Context Protocol ApplicationsLLM agents and tool use → Agent threats, guardrails and governance
SkillGate: Cost Efficient Runtime Malicious Skill File Detection in Coding AgentsLLM agents and tool use → Agent threats, guardrails and governance
MemSFT: Mitigating Alignment Tax with an External Parametric MemoryLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Beyond Self-Knowledge: Propagating Uncertainty Across Reasoning and Retrieval in LLMsLLM reasoning → Knowledge- and tool-augmented reasoning +1
Forensic Reproducibility Audit of a Radiology Vision-Language Model Benchmark: From Intended Protocol to Released ArtifactMultimodal LLMs / vision-language models → Medical vision-language models
The LAIA Dataset: Labelled Attention for Intelligent AutomobilesEnd-to-end autonomous driving → VLMs for autonomous driving
ARCHER: Agentic Rule and Compliance Harness for Executable RegulationsLLM agents and tool use → Agent threats, guardrails and governance
Distilling Temporal Search and Reasoning: Evolving LLMs for Future Prediction via Harness-Assisted Efficient Data SynthesisLLM reasoning → Time series reasoning +1
OrthKD: Extracting Generalized Clinical Knowledge from Heterogeneous Teachers for Lightweight DeploymentKnowledge distillation → Distillation for medical imaging
A Causality-aware Infer-diagnose-refine Framework for Test-time Modality Adaptation in VLA ModelsVision-language-action models → Efficient VLA inference and adaptation
Group Equivariant Diffusion for Anomaly Detection in Computational CytologyDiffusion models → Medical imaging and segmentation
Beyond Prefill-Decode Disaggregation: Dissecting LLM Inference for Heterogeneous Platforms via Dynamic Operator SchedulingEfficient LLM inference and serving → Serving systems and request scheduling
Agentic AI in medicine: architectures, applications, evaluation, and challenges for clinical translationLLM agents and tool use → Medical and scientific domain agents
CoTinyVLA: Chain-of-Thought Distillation for a Sub-Billion-Parameter Vision-Language-Action ModelLLM reasoning → Spatial and embodied reasoning +1
PatientAgentBench: A Benchmark Framework for Evaluating Patient-Facing Health AI AgentsLLM agents and tool use → Medical and scientific domain agents

All 414