Papers

New on arXiv

Sun 16 Aug 2026

226 papers

Improving LLM-based Autonomous Web Agents with FilteringLLM agents and tool use → Web, search, and research agents
Alignment of LRMs via Counter-Aligned Few-Shot Conversation ExposureLLM reasoning → Reasoning for agents, planning and code +1
Self-Indexing Attention for Compression-Compatible Sparse Long-Context LLM InferenceEfficient LLM inference and serving → KV cache optimization
Aligning Human Sense: Calibrated Distributional Reward Learning for Video GenerationLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
Agentic Security: A Systematization of Tools, Failure Modes, and Design Laws for LLM-Driven Penetration TestingLLM agents and tool use → Agent threats, guardrails and governance
Adversarial Review: Structured Disagreement for Grounded Agentic Code ReviewLLM agents and tool use → Software engineering coding agents
TractoGraphVLM: A Unified Vision-Language Framework for White Matter TractographyMultimodal LLMs / vision-language models → Medical vision-language models
The Price of Thinking: Reasoning Effort as a Model-Specific API ContractLLM reasoning → Test-time compute and efficient thinking
WONDER: A Radio World Model-based Negotiation Framework for Multi-Agent UAV Coverage OptimizationWorld models → World models for navigation
The Little Scientist: LLM Agent-Driven Discovery via the Scientific MethodLLM agents and tool use → Web, search, and research agents
SEER: Long-Context Reasoning via Selective Visual-Text CompressionLLM reasoning → Multimodal reasoning
Solvable Sokoban Without a Solver via DiffusionDiffusion models → Discrete and text diffusion
Navigation-Informed Embeddings: Dense-Retriever Adaptation from Agent Search TracesLLM agents and tool use → Web, search, and research agents
KV-Pipe: On the Relation Between KV Sharing and Pipeline Parallel Efficiency in LLMsEfficient LLM inference and serving → KV cache optimization
Aborted but Not Forgotten: KV-Cache Retention Breaks Rollback Consistency in Language AgentsEfficient LLM inference and serving → Efficient reasoning and agent inference
Revisiting Open-Loop Execution in Robotics: Toward Reactive, Higher-Performing PoliciesOffline RL and imitation learning from data → Robot manipulation imitation learning
UI-Mate: Advancing Open-Weight Foundation GUI Agents with In-Context DemonstrationsLLM agents and tool use → GUI and mobile agents
Unified Pedestrian Path Prediction Using Inverse Reinforcement LearningDeep reinforcement learning methods → RL for autonomous driving and traffic
RAPAC-DP: Response-Aligned Pending-Action Compensation for Diffusion Policies under Delayed ExecutionDiffusion models → Diffusion and flow policies for robotics
Noesis: Bidirectional Graph-RAG with Adaptive Parallelism and Cross-Knowledge-Base Semantic DiscoveryRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
Crystal-structure design by agentic AI in a language of motifsLLM agents and tool use → Medical and scientific domain agents
Bounded Agents: Delegation Security for Multi-Agent AI SystemsLLM agents and tool use → Agent threats, guardrails and governance
Deploying Frontier Agentic Technology in MOOSEnger, a Multiphysics-Capable AI AssistantLLM agents and tool use → Multimodal and embodied agents
Dear Algo: A Precision-First Agentic Intent Layer for Unified Search and RecommendationLLM agents and tool use → Web, search, and research agents
GigaBrain-0.7: Scaling Embodied Foundation Models to Emergent Capabilities with a Three-System ArchitectureVision-language-action models → Generalist VLA models and benchmarks
Beyond Visual CoT: Internalized Visual Thinking for Proactive Video ReasoningLLM reasoning → Multimodal reasoning +1
RAGas: Retrieval-Augmented Gas Optimization for Smart Contracts with Continuous Knowledge IntegrationRetrieval-augmented generation → Privacy and security of RAG
Query Expansion Should Be Coordinated: Dense Expands, Sparse AnchorsRetrieval-augmented generation → Generation and context integration
MicroVerse: An Instrument for Measuring Self-Authored Identity Drift in Long-Horizon Multi-Agent Language-Model SimulationsLLM agents and tool use → Social simulation and multi-agent debate
Self-Supervised Auxiliary Task Discovery for Stable Reinforcement Learning in Stock TradingDeep reinforcement learning methods → Adversarial robustness and RL in finance
Schema-Agnostic Graph Reasoning Agent for Hybrid Knowledge GraphsLLM reasoning → Knowledge- and tool-augmented reasoning +1
RLCascadeRouter: Quality-Estimator-Free Cascade Routing via Reinforcement LearningDeep reinforcement learning methods → RL training of language and agent models
ViTaR: Visuo-Tactile Residual Adaptation for Foundation VLA ManipulationVision-language-action models → Tactile, force-aware and humanoid VLA
KV-Rescue: Recovering Reasoning Language Model KV Eviction Loss via Stepwise InterleavingLLM reasoning → Test-time compute and efficient thinking +1
Routing Divergence Is Not Evidence of Behavioral Influence in Same-Weight MoE Self-DistillationMixture-of-experts → Expert routing strategies +1
ALKEMIE Agent: an autonomous platform for computational materials designLLM agents and tool use → Medical and scientific domain agents
Learning Stock Trading Policies via Barycenter-Based Adversarial Inverse Reinforcement LearningOffline RL and imitation learning from data → Inverse reinforcement learning +1
Training Agents to Evolve with Their Harness: TaoLive Digital Avatar Agent Technical ReportLLM agents and tool use → Software engineering coding agents
Intent-Driven Situation Tracking for User-Centric Multi-Turn AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
Making two action heads agree: coordination mechanisms and a runtime collapse certificate for flow-matching policiesDiffusion models → Diffusion and flow policies for robotics
Toward AI-Friendly Cartography: Understanding How Color Design Influences Foundation Model Spatial Reasoning on Sequential Choropleth MapsLLM reasoning → Spatial and embodied reasoning
FirstDiff: One-Step Diffusion-Based Anomaly Detection for Multivariate Time Series via Initial Noise PredictionDiffusion models → Flow matching, time series and science
Continuous Quantum Feedback Control via Kraus-Parameterized Belief Reinforcement LearningDeep reinforcement learning methods → Quantum reinforcement learning
GAINS: Leveraging Inconsistent Human Intervention Signals in Reinforcement LearningDeep reinforcement learning methods → RL for robot locomotion and manipulation
PixelControl: Fine-Grained Condition Fidelity in Text-to-Image DiffusionDiffusion models → Image generation, text-to-image and editing
HyMem: Hierarchical Context Management for Long-Horizon Agents via Information IsolationLLM agents and tool use → Agent memory systems
Adaptive Mixing of Policies from Searching and Policies from LearningDeep reinforcement learning methods → Deep RL algorithms and representations
ConceptFormer: Learning Adaptive Latent Concepts for Query-Document Alignment in Visual Document RetrievalRetrieval-augmented generation → Multimodal RAG
THESIS-MoE: Trainable Hierarchical Extraction and SteerIng of Sycophancy in Mixture-of-ExpertsMixture-of-experts → MoE large language models
Robo-Dopamine 2.0: History-Conditioned and OOD-Aware Process Reward Modeling for Robotic ManipulationLLM reasoning → Spatial and embodied reasoning +3
Where Accountability Lives: Mapping Human Responsibility to Workflow Artifacts in Agentic Software DevelopmentLLM agents and tool use → Software engineering coding agents
PL-Guard: Probabilistic Logic Reasoning for LLM GuardrailsLLM reasoning → Reasoning for agents, planning and code +1
Adaptive Heterogeneous Compression for Resource-Efficient Federated Knowledge DistillationKnowledge distillation → Federated, robust and private distillation
When Stories Evolve: Benchmarking LLM Storytelling Across Agent Architectures in Open-Ended World SimulationsLLM agents and tool use → Multimodal and embodied agents
Algorithm-Architecture Co-Design for Efficient VLA Inference via Speculative Inference and VerificationVision-language-action models → Efficient VLA inference and adaptation +1
EgoGazeLite: On-Device Egocentric Gaze Prediction for Token-Efficient Multimodal LLM Video InputMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
AlloEgo-VLM: Disambiguating Allocentric and Egocentric Reference Frames in Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
VARM-Bench: Benchmarking Verifiable Structured Reasoning in Chinese Abusive Speech ModerationLLM reasoning → Reasoning benchmarks and evaluation
TRACE: Trajectory Aware Reasoning for Multi-Turn Adversarial Conversation EvaluationLLM reasoning → Reasoning for agents, planning and code +1
When Entropy Is Not Enough: Reclaiming Lost Semantics in LLM Output Length PredictionEfficient LLM inference and serving → Serving systems and request scheduling
Agent Gym: A Framework for Continuous Evaluation and Evolution of LLM Agents Through Human-in-the-Loop FeedbackLLM agents and tool use → General LLM agent frameworks and benchmarks
GraniKV: Asymmetric Granularity KV-Cache Paging for Multi-Agent Systems with Long Shared PrefixEfficient LLM inference and serving → Efficient reasoning and agent inference
PoseAdapter: Dual-Stream 2.5D Controllable Image Generation for Complex Multi-Object ScenesDiffusion models → Image generation, text-to-image and editing
Kozuchi Agent: A Language-Agnostic Open-Weight Agent for Software RepairLLM agents and tool use → Software engineering coding agents
ARENA: Automated Red-Teaming for Large Audio Language ModelsLLM safety and security → Audio jailbreaks and speech backdoors
Catching Hallucinated Citations in Video-LLM Question Answering: A Self-Verification Pipeline and Verifier Ablation StudyMultimodal LLMs / vision-language models → Video understanding with MLLMs
Not All History Helps: Velocity-Aware Selective Memory for Long-Horizon End-to-End Autonomous DrivingEnd-to-end autonomous driving → Generative simulation for driving
MistyPilot: Enabling Social-Robot Control through Multi-Agent LLM Skill OrchestrationLLM agents and tool use → General LLM agent frameworks and benchmarks
Spectral Saliency for Machine UnlearningLLM safety and security → Machine unlearning
BengaliMCQ: Automatic Generation and Answer Prediction of Academic Multiple-Choice Questions in a Low-Resource LanguageRetrieval-augmented generation → RAG robustness and adaptive retrieval
CrossView: Can Vision-Language Models Reason Across Cameras?LLM reasoning → Multimodal reasoning +1
DeltaLog: Deferred Materialization of Recurrent States for Linear Attention DecodingEfficient LLM inference and serving → KV cache optimization
Why Summaries Turn Neutral: Policy Attribution for Sentiment Drift in Reinforcement Learning from Human FeedbackLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
GLaQ: Grounding Latent Queries in Visual Evidence for Multimodal ReasoningLLM reasoning → Multimodal reasoning +1
UniFed-VLM: Federated Instruction Tuning for Vision-Language Models with Multiple HeterogeneityMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Temporal Logic Guided Universal Task Representations for Reinforcement LearningDeep reinforcement learning methods → Interpretable and logic-guided RL
Do Language Models Consistently Encode the Current Year?LLM reasoning → Knowledge- and tool-augmented reasoning
EcoVLA: Energy-Efficient Device-Edge Co-Inference for Vision-Language-Action Models under Real-Time ConstraintsVision-language-action models → Efficient VLA inference and adaptation
HxAgent: Iterative Agent Planning for End-to-End Web Application TestingLLM agents and tool use → Web, search, and research agents
Q-First: Most of Attention Needs Only the Query in Disaggregated LLM DecodingEfficient LLM inference and serving → KV cache optimization
Not All Attention Is Equal: A Quantitative Survey of the EEI Trade-offLLM reasoning → Multimodal reasoning
Spatially-Grounded Flow Matching: Structured Source Distributions for Image GenerationDiffusion models → Flow matching, time series and science
"What's That Sound?": A Versatile, Robust, and Lightweight Convolutional Transformer for Environment Sound Recognition
Correcting Learning-based Perception for Safety
Fusion Anything: A Generalized Multimodal Foundation Model
PRQuant: Permutation Residual Quantization for Low-Overhead Inference
How Language Models Choose Sides: Internal Representations of Instruction Hierarchy
Self-Specialized Teachers for Domain Post-Training
Few-Shot Cross-Dataset Adaptation for Tuberculosis Detection Using DenseNet
AI Visual Inspection for Garment Production
EditStream: A Unified Autoregressive Framework for Interactive Video Generation and Editing
Demons on a Budget: Adaptive Measurement Placement at the Entanglement Phase Transition
RDFdL: Integrating RDF with Differential Dynamic Logic
MagViT: Interpretable Multi-Magnification Transformers with Patient-Level Model Selection for Breast Histopathology
ORViT-DR: Ordinally-Robust Hybrid ViT for Low-Resolution Diabetic Retinopathy Grading
CM-MAE: A Physics-Guided Cross-Modal Self-Supervised Learning Framework for Vision-Wireless Applications
The Limits of Binding in Dual Encoders
BagShift: Measuring How Patch Selection Changes the Evidence Seen by Whole-Slide MIL
A Banach-Space Theory of Markovian Halpern Iteration for Non-Expansive Maps
Beat the Counter First: A Baseline for Temporal-Graph Anomaly Detectors

All 226