Papers

New on arXiv

Sun 20 Sep 2026

244 papers

Query Independent Variable Rate Visual Token CodingEfficient LLM inference and serving → Visual token pruning in multimodal models
PrimeSeeker: Capability-Oriented Supervision for Deep Search AgentsLLM agents and tool use → Web, search, and research agents
Local Predictability and Collective Fidelity in LLM-Agent SocietiesLLM agents and tool use → Social simulation and multi-agent debate
Lookahead-R: Budget-Aware Tool Retrieval via Execution-Centric PlanningLLM agents and tool use → Tool learning and function calling
When Successful Memories Mislead Embodied Agents:Memory Adaption For Task-Conditioned ExecutionLLM agents and tool use → Agent memory systems
Can't Find Waldo: Evaluating VLMs' Sensitivity to Image Resolution and Detail LevelMultimodal LLMs / vision-language models → Visual token pruning and efficiency
What Makes a Terminal-Bench Task Hard? Separating Genuine Hardness from Fake-Hardness on an Adjudicated Agentic CorpusLLM agents and tool use → Software engineering coding agents
MAC-RRG: Iterative Multi-Agent Collaboration for X-ray Radiology Report GenerationLLM agents and tool use → Medical and scientific domain agents
Impact Is Not Invalidation: Ask About the Claim, Not the DiffLLM agents and tool use → Agent memory systems
Agents That Edit Documents: Measuring Agentic PDF Forgery Against a Non-Agentic ControlLLM agents and tool use → Agent threats, guardrails and governance
Topology-Informed Visual Prompting For Vision Language Action PoliciesVision-language-action models → 3D and spatial-aware VLA
ReVeal: A Reconstruction-Aware Real-to-Sim Framework for VLA Policy EvaluationVision-language-action models → Failure recovery and robustness
LumoTree: Path-Parallel Speculative Verification for Hybrid Language ModelsEfficient LLM inference and serving → Speculative decoding
Connecting the Dots in Agentic AI Security: A Cross-Dimensional Threat Taxonomy, Evaluation Maturity, and Open ChallengesLLM agents and tool use → Agent threats, guardrails and governance
Circuit-Diff: Factual Edit-based Intervention Method for Localizing Knowledge in Attribution GraphsMechanistic interpretability → Circuits and mechanistic analysis
MotionJEPA: Preventing Temporal Feature Collapse by Capturing Visual Changes in Latent SpaceWorld models → Latent predictive world models
Q-TIE: A Lightweight and Generalizable Re-ranking Framework for Temporal Information RetrievalRetrieval-augmented generation → Generation and context integration
Grounded Action Model: 3D Grounding as a Foundation for RoboticsWorld models → Embodied and robot world models +1
Pretraining of Medical Visual Encoders Toward Multi-modal Large Language ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Vibe-GUIDE: A Graph-based User Interface in IDEs for Oversight in Vibe CodingLLM agents and tool use → Software engineering coding agents
Structured World-State Reasoning for Agentic Robotic SearchLLM agents and tool use → Multimodal and embodied agents
SPLASH: Co-Designing Sparse Attention with High-Bandwidth Flash for Efficient Long-Context InferenceEfficient LLM inference and serving → KV cache optimization
Confidence-Aware Teacher-Student Distillation for 3D Medical SegmentationKnowledge distillation → Distillation for medical imaging
FLARE: A Full-Lifecycle Dense Supervision Paradigm for Long-Horizon Coding Agents via Generative Reward ModelLLM reasoning → Reasoning for agents, planning and code +2
Total Cost of Agency: Exact Attribution of Memory Injection Cost in Multi-Agent LLM WorkflowsLLM agents and tool use → Software engineering coding agents
Statistical Convergence of Transformer Encoder-Accelerated Robust Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
PRISM-RAG: Multimodal Hypergraph Retrieval-Augmented Generation for Tobacco Product and Legislative Policy ReasoningRetrieval-augmented generation → Multimodal RAG
OnlineWM: Causality-Aware Active Online Learning for Effective World ModelingWorld models → Latent predictive world models
Constrained Decoding Eliminates Structural Failures in Small LLMs but Reveals a Scale-Dependent Semantic GapLLM agents and tool use → Tool learning and function calling
GRACE: Grounded Adversarial Reasoning over Canadian LawLLM reasoning → Reasoning benchmarks and evaluation
Layer-Aware Position Embeddings for Visual Token Pruning in Multimodal Large Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
When the Agent Becomes the Kernel: A Systematization of Security on the Path to AI-Native Operating SystemsLLM agents and tool use → Agent threats, guardrails and governance
Distill What You Trust: Reliability-Aware Multi-Teacher On-Policy DistillationKnowledge distillation → On-policy and self-distillation for LLMs
PhysAI-Bench: A Benchmark for LLM-Based Agentic Decision-Making in Autonomous UAV-Centric Physical AILLM agents and tool use → General LLM agent frameworks and benchmarks
Why Do Video Diffusion Models Violate Physics? Unveiling the Flaws in Attention MechanismsDiffusion models → Video generation
WOLF: World Model Guided LiDAR Exploration with Predictive FrontiersWorld models → World models for navigation
Reassessing Global Gradient-Norm Imbalance in BLIP Fine-Tuning Across Physical DomainsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Beyond Appearance Shifts: Task-Semantic Action Calibration for VLA ModelsVision-language-action models → Failure recovery and robustness
Spiking Neural Network Actor-Critic Proximal Policy Optimization Control for Autonomous UAV Navigation Through Constrained Openings in Civil Infrastructure and BuildingsDeep reinforcement learning methods → RL for robot navigation and UAV control
Are Human-Aligned Models Models of Humans? A Turing-Test Gap in Preference AlignmentLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
CompVLA: A Variable Compliance Vision-Language-Action Model for Contact-rich ManipulationVision-language-action models → Tactile, force-aware and humanoid VLA
PETR: Prompt Ensembling with Training-free Routing for Vision-Language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Cost-Aware Reinforcement Learning with Action Masking and Projection for Battery Energy Storage Dispatch under Suppressed-Spread Market ShiftsDeep reinforcement learning methods → RL for energy, networks and operations
Listen Then Reason: Perception-Grounded Test-Time Reinforcement Learning for Large Audio-Language ModelsLLM reasoning → Audio-language model reasoning
On the Efficiency-Safety Dilemma in Large Reasoning ModelsLLM reasoning → Reasoning for agents, planning and code +1
An Efficient and Effective Watermarking Scheme for the Protection of the Intellectual Property Rights of Video Generative ModelsLLM safety and security → Watermarking generated content and models
AR-WAM: A Visual-Conditioned Agent-Ready World Action Model for Robotic ManipulationWorld models → Embodied and robot world models +1
VibeMemBench: Evaluating Memory Systems for Coding Agents on Real Repository Coding TasksLLM agents and tool use → Software engineering coding agents
MaskVLA: Visual Masking Against Trajectory Overfitting of Vision-Language-Action ModelVision-language-action models → Generalist VLA models and benchmarks
Transferring Visual Explanations: How Cross-Architecture Knowledge Distillation Affects Model InterpretabilityKnowledge distillation → General knowledge distillation methods
AgentBetta: Verification-Driven Adaptive Configuration of an AI Nano-Agent through Selective Expansion and Verified ContractionLLM agents and tool use → General LLM agent frameworks and benchmarks
Stability-Aware Imitation Learning from Model Predictive Control for Autonomous Vehicle Lateral Control: Exact Q-Loss and a Novel Training ProcedureEnd-to-end autonomous driving → Imitation-learning end-to-end planners +1
Runtime Authorization Consistency Checking for MCP-based Agentic WorkflowsLLM agents and tool use → Agent threats, guardrails and governance
BabelArena: A Large-Scale Multilingual Benchmark for LLM AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
FeasibleFlow: One-Step Joint Transport of Configuration Feasibility and Trajectories for End-to-End DrivingEnd-to-end autonomous driving → End-to-end system architecture
Cognitive Action Reasoning for Proactive Robots from Human-Centered Multimodal ObservationsLLM reasoning → Spatial and embodied reasoning
STRIDER: Stepping-Enabled Multi-Gait Hierarchical 3D Loco-Manipulation Framework for Humanoid RobotsDeep reinforcement learning methods → RL for robot locomotion and manipulation
Generative Learning for Ambisonic UpscalingDiffusion models → Audio, speech and music
RPMem: Learning Long-Term Recurrent Parametric Memory Across Sessions for LLM AgentsLLM agents and tool use → Agent memory systems
Propose, Verify, Commit: Evidence-Grounded Memory for Long-Horizon Multi-Actor ConversationsLLM agents and tool use → Agent memory systems
PSD: Pseudo Self-Distillation of Memory Representation Capabilities for LLM AgentsLLM agents and tool use → Agent memory systems +1
PhysReflect: Geometry and Perception Guided Diffusion for Physically-Plausible Mirror ReflectionsDiffusion models → 3D, point cloud and scene generation
Tool-Augmented On-Policy Distillation for LLM Domain Adaptation in Sequence-Based Omics TasksKnowledge distillation → Knowledge distillation for LLMs
Bayesian Filtering in Physical Systems via Test-time Trained Flow MatchingDiffusion models → Flow matching, time series and science
One to More, More to One: Category-Aware Iterative Expert Training for Software Engineering AgentsDeep reinforcement learning methods → RL training of language and agent models
If You Hear It, Help Find It: Orthogonal Knowledge Distillation for Open-Vocabulary Audio-Visual Event LocalizationKnowledge distillation → Speech and audio distillation
The Evidence Ladder for Reinforcement Learning in Healthcare: From Retrospective Policies to Trusted InterventionsDeep reinforcement learning methods → RL for clinical treatment and surgery
The Right Future for Action: Learning Action-Relevant Predictive States in World Action ModelsWorld models → Embodied and robot world models
TicTacBench: Benchmarking Timing Closure Capabilities of Coding AgentsLLM agents and tool use → Software engineering coding agents
MinCU: A Fine-Grained Benchmark for Grounded Minimal-Change Understanding in Image PairsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Graph Memory for LLM Agents: At What Cost? A Comparative Evaluation of Query, Ingest, and Update Performance Across Graph Database EnginesLLM agents and tool use → Web, search, and research agents
ValueDiff: Value-Geometric KV Cache Eviction for Sink-Suppressed LLMsEfficient LLM inference and serving → KV cache optimization
SCULPT-VLA: Learning Structured Control through Staged Action GroundingVision-language-action models → Human-video and latent-action pretraining
Transferability and operational reliability of a Prithvi crop classification foundation model under phenological and geographic shift across three continents
Dynamical low-rank equilibrium computation for stochastic games between advanced persistent threats and moving target defense
ShatterQuant: Breaking Uniform Precision with Block-Wise Mixed-Precision on a Systolic Transformer Hardware Accelerator
From Web(logs) to Web(AI): Questions, Platforms, and Methods across Twenty Editions of ICWSM
When a Data Artifact Isn't a Shortcut: Causal Auditing of Synthetic RLVR Corpora
How to Run Statistics over LLM Judges and Trust the Results: Calibrated Inference for Small-Sample AI Evaluation with evalstats
Constructing Challenging Browser-Use Tasks by Controlled Environment Interventions
Automated Evaluation of Multi-Turn Dialogues in In-Car Conversational Assistants
TRACE: Deployable Tree-Relational Structure Enhancement for Oncology LLMs
Can Multimodal Large Language Models Generate and Detect Multimodal Social Media Fake News?
Cross-Dataset Generalization of Bangladeshi Rice Leaf Disease Classifiers: Benchmark, Diagnosis, and Mitigation
MDL-Calibrated Significance-Gain Pair Encoding: Replication-Aware Automatic Stopping for Subword Tokenization
Does DCGAN-Based Synthetic Augmentation Improve Brain Tumor MRI Classification? An Empirical Study
Bridging LLM Serving and CXL-SSDs with Chunk-Aware KV Cache Management
Text Scores Do Not Establish Performance on Lexically Non-Diagnostic Speech Tasks: A Qwen2-Audio Quantization Case Study
HARN: Hierarchical Associative Resonance Network for Event-Driven Multi-Timeframe Forecasting
Signal2Symbol: Neuro-Symbolic Temporal Reasoning for Explainable Physiological Time-Series Anomaly Detection
The Probabilistic Structure of Large Language Models
Entropy Can Flow, or It Can Guide. Be Entropy. LEDFlow: Introducing Entropy-guided Generation Order into Uniform Discrete Flow
WILSON - a pathology foundation model framework for patient-level analysis and diagnostic text generation
Rachel: A general-purpose language model directs and revises retrosynthetic routes
You've Seen Enough: Quality-Constrained Image Coding for Machines
HaikuS2S: A Cascaded System For Responding In Verse
ORION-CMR: On-scanner Reporting with Integrated Foundation Model for End-to-End Cardiac MRI Analysis and Interpretation
Echo State Network (ESN) for Signal Recovery in RF-Impaired IBFD MIMO Systems
XYEval: Agents say yes to bad advice
Sparse Regression Distilled from a Single Robust Fit

All 244