Papers

New on arXiv

Fri 31 Jul 2026

411 papers

LLM-Guided Transformation of Non-Critical Driving Scenes into Safety-Critical Scenarios Using Augmented RealityEnd-to-end autonomous driving → VLMs for autonomous driving
Z-Loss Backward Geometry in Dense Output Heads and Sparse RoutersMixture-of-experts → Expert routing strategies
Diagnosing and Mitigating Perception-Decision Misalignment in Omni-LLMs via Modality Subspace ActivationWorld models → World models for language agents +1
Measuring Cross-Task Behavioral Consistency in Language Model AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
Secret-Stego Dissimilarity as a Design Axis: Invertible Coverless Image Steganography with Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects
Decode-Branch Transformers: Decoupling the Primary Prefill Path from Additional Decode ComputationEfficient LLM inference and serving → Early exit and edge inference
RecSys Factory: Bounding LLM Agent Autonomy to Decision Points in the Industrial Recommender LifecycleLLM agents and tool use → Agentic recommender systems
Towards Query-Agnostic RAG Evaluation via Query Coverage and Claim VerifiabilityRetrieval-augmented generation → RAG robustness and adaptive retrieval
TRACE Bench: Task-driven Roleplay Agentic Checklist EvaluationLLM agents and tool use → Social simulation and multi-agent debate
CORA-Diff: Confidence-Oriented Residual Acceptance for Efficient Diffusion Language Model InferenceDiffusion models → Discrete and text diffusion
Backtrader-Bench: Benchmarking LLM Agents on Algorithmic Trading with Self-Generated MCQsLLM agents and tool use → Financial trading and forecasting agents +1
LinearKV: One Cached State Suffices for Position-Independent Caching in Hybrid LLMsEfficient LLM inference and serving → KV cache optimization
Generalizing deep reinforcement learning across cable-driven parallel robot configurations with actuator-level policiesDeep reinforcement learning methods → RL for robot locomotion and manipulation
DarwinX: Evolving Agent Harnesses Through Natural SelectionLLM agents and tool use → General LLM agent frameworks and benchmarks
Beyond Routing Weights: Faithful Response-Level Interpretation of Mixture-of-Experts Reward Models via Contribution ContrastLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
EntropyMoE: Entropy-Aware Sparse Expert Routing for Tokenizer-Free LLMsMixture-of-experts → Expert routing strategies
Agentic Planning for Symbolic ExecutionLLM agents and tool use → Software engineering coding agents
TEXAS: Task-Expert-Aware Supervision for Downstream Mixture-of-Experts LLM AdaptationMixture-of-experts → Expert routing strategies
Position: It's Time to Optimize LLMs for Self-ConsistencyLLM reasoning → Test-time compute and efficient thinking
HyperAgent: Planning and Acting over Tool-Schema Hypergraphs for Tool-Use LLM AgentsLLM agents and tool use → Tool learning and function calling
Verified Tool Calls Improve LLM Agent Reliability Under Non-Atomic FailuresLLM agents and tool use → Tool learning and function calling
CUADebug: Diagnosing and Repairing Computer-Use Agent FailuresLLM agents and tool use → GUI and mobile agents
MDArena: Evaluating Coding Agents on Realistic Molecular Dynamics WorkflowsLLM agents and tool use → Medical and scientific domain agents
Rethinking Self-Evolving Agent Skills: Feedback Dynamics over Multiple RoundsLLM agents and tool use → General LLM agent frameworks and benchmarks +1
CurveShift: Is Agent Progress Scalar? Separating Level from ShapeLLM agents and tool use → General LLM agent frameworks and benchmarks
ORCA: ORgan-Centroid Aggregation for Training-Free 3D CT Visual Token CompressionMultimodal LLMs / vision-language models → Medical vision-language models +1
RMSWeb: Reflection, Failure-Mode Mining, and Salvage-DS for Web Agent Reinforcement LearningLLM agents and tool use → Web, search, and research agents +1
Learning to Coordinate Symbolic Tools: LLM Agents for Verified Sum-of-Squares CertificatesLLM agents and tool use → Educational agents and math problem solving
Agentic Bayesian Optimization through Surrogate-Augmented AutoresearchLLM agents and tool use → General LLM agent frameworks and benchmarks
Towards General Language-Conditioned Latent Safety FiltersVision-language-action models → Safety and adversarial attacks
CrystalMem: Elastic Memory for Self-Evolving LLM Agents via Knowledge CrystallizationLLM agents and tool use → Agent memory systems +1
Abstention as an Action Can Kill Both the Reward Gradient and the KL Anchor: Collapse Law and Repair for Error-Penalized Reinforcement LearningDeep reinforcement learning methods → RL training of language and agent models
WM-Cov: Test Adequacy for Interactive World-Model-Style Autonomous Driving SimulationWorld models → World models for autonomous driving +1
Learning to See Locally and Align Clinically with Pathology Semantics for Radiology Report GenerationMultimodal LLMs / vision-language models → Medical vision-language models
LoopsBench: From Harness Engineering to Loop Engineering in Coding Agent EvaluationLLM agents and tool use → Software engineering coding agents
Cross-Task Dissociation in Frontier Vision-Language Model Theory of MindMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Similarity-Aware Machine UnlearningLLM safety and security → Machine unlearning
More Debate, Same Evidence: Structural Limits of Homogeneous Multi-Agent GroundednessLLM agents and tool use → Social simulation and multi-agent debate
Latent-Centroid Steering: Single-Pass Classifier-Free Guidance for Command-Aligned Autonomous DrivingMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection +1
Attention-Steered Vision-Language Models for Sign Language TranslationMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Learning How Much, Not Just What: Cross-Patient Burden Order for CT Vision-Language PretrainingMultimodal LLMs / vision-language models → Medical vision-language models
A Few Neurons Reveal When LLMs Misuse Tools: Sparse Detection and Selective Steering for Reliable Tool UseLLM agents and tool use → Tool learning and function calling
Personalizing Large Language Model Agents with Small Policy ModelsLLM agents and tool use → Agent memory systems
From Monoliths to Swarms: A Study of Attack Surface Evolution in the Transition to Multi-Agent Web SystemsLLM agents and tool use → Agent threats, guardrails and governance
TRACE-TS: Attribution-Grounded and Traceable Sensor-Language Reasoning for Human Activity UnderstandingLLM reasoning → Spatial and embodied reasoning
Cross-Benchmark Generalization in Long-Horizon AgentsDeep reinforcement learning methods → RL training of language and agent models
AgentStream: How Well Do Self-Evolving LLM Agents Perform Under Streaming Tasks?LLM agents and tool use → General LLM agent frameworks and benchmarks +1
RadPRISM: Schema-stratified radiology-report supervision for concept-disentangled image representations and visual groundingMultimodal LLMs / vision-language models → Medical vision-language models
DiffusionGemma Technical ReportDiffusion models → Discrete and text diffusion
Leak It: Per-Document Extraction Beyond Aggregate Membership InferenceLLM safety and security → Membership inference attacks
Symbolic Attack Chain Generation from Atomic Red Team Techniques: An Empirical Study of Predicate Representation GranularityLLM safety and security → Prompt injection and agent security
Stateful Cooperative Agents Safeguarding LLMs Against Evolving Multi-Turn AttacksLLM agents and tool use → Agent threats, guardrails and governance
Progressive2: A Teacher-Student Progressive Co-Evolving Knowledge Distillation Method for Substantial Model CompressionKnowledge distillation → General knowledge distillation methods
Shared Organizational Memory for Enterprise Coding Agents: System Design and Deployment SnapshotLLM agents and tool use → Agent memory systems
Track-Guided Hierarchical Reinforcement Learning for Autonomous Vehicle Drifting with Minimum-Lap-Time PlanningEnd-to-end autonomous driving → Reinforcement learning for driving +1
Distill What RGB Can Recover: Privileged 3D Evidence for RGB-Only Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
MetaRoute-Bench: Evaluating Meta-Decision Policies for Agentic Workflow RoutingLLM agents and tool use → General LLM agent frameworks and benchmarks
Learning Compositional Meta-Routing for Agentic Workflows: An Executable BenchmarkLLM agents and tool use → General LLM agent frameworks and benchmarks
Skillsets on the Chain: A Blockchain-based Zero-Trust Framework for Agentic AI NetworkingLLM agents and tool use → Agent threats, guardrails and governance
ExtractBench: A Benchmark for Schema-Guided Enterprise Document ExtractionLLM agents and tool use → Web, search, and research agents
Reusing Past Repairs Through Hierarchical Trajectory Abstraction for Coding AgentsLLM agents and tool use → Software engineering coding agents
CodeShrink: Adaptive Visual Compression for Efficient Multimodal Code UnderstandingMultimodal LLMs / vision-language models → Visual token pruning and efficiency
AgentHPOBench: A Benchmark For Evaluating LLM Agents as Sequential Hyperparameter OptimizersLLM agents and tool use → General LLM agent frameworks and benchmarks
RayViT: Ray-Conditioned Visual Representations for Viewpoint-Robust Imitation LearningOffline RL and imitation learning from data → Robot manipulation imitation learning
When Does On-Policy Interaction Help? Representational Tradeoffs in Value-Based Imitation LearningOffline RL and imitation learning from data → Imitation learning methods
WCM: A World Critic Model for Vision-Language-Action Reinforcement LearningVision-language-action models → Reinforcement learning and post-training +1
Educating the Agentic Engineer: Curricula, Collaboration, and Continuous Learning in the AI EraLLM agents and tool use → Software engineering coding agents
FriendBench: Benchmarking Dyadic Familiarity Inference in Humans and Multimodal Large Language ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
HAM-VLN: Harnessing Hierarchical Agentic Memory for Zero-Shot Vision-and-Language NavigationLLM agents and tool use → Multimodal and embodied agents
FibVLA: An Efficient Temporal Vision-Language-Action Model with Fibonacci SamplingVision-language-action models → World model-based VLA
Convergence and Regret of the Policy Gradient for Multi-Armed Bandits in Diffusion EnvironmentDeep reinforcement learning methods → RL theory and sample complexity
ResKV: Reconstructing Omitted Attention Contributions for Fixed-Budget KV Cache CompressionEfficient LLM inference and serving → KV cache optimization
SLIM: Saturation-Aware Lightweight Performance Modeling for LLM ServingEfficient LLM inference and serving → Serving systems and request scheduling
Safe Vision Language Action Models via Barrier Enhanced Flow MatchingVision-language-action models → Safety and adversarial attacks +1
LEMUR: Learning to Align with Multi-Objective Reinforcement Learning from Preference FeedbackLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
AMTFV: Agentic Mathematical Tool-Flow Verification for LLM Self-CorrectionLLM agents and tool use → Educational agents and math problem solving
STAGE: STyle-controllable Action GEneration for personalized autonomous drivingEnd-to-end autonomous driving → Imitation-learning end-to-end planners
From Code Review to Code Critique: Intent, Drift, and Spotlight for AI-Generated Diffs at ScaleLLM agents and tool use → Software engineering coding agents
Adaptive FastOPD: Progress-Aware Rollout Horizon Expansion for Efficient On-Policy DistillationKnowledge distillation → On-policy and self-distillation for LLMs
DreamQAS: Learning a Decision-Useful World Model for VQE-Efficient Quantum Architecture SearchWorld models → Latent dynamics models and model-based RL
Evidence-Type Competition: When Can Interventional Data Teach Language Models Causal Direction?LLM reasoning → Causal reasoning in LLMs
Temporal Policy: History-Initialized Action Generation for Robotic Learning from DemonstrationDiffusion models → Diffusion and flow policies for robotics +1
Weight-Space Mixture-of-Experts for Implicit Neural Representation ClassificationMixture-of-experts → Vision mixture-of-experts
MoPET: Parameter-Efficient Mixture-of-Experts for Unified Medical Image ClassificationMixture-of-experts → Medical and biomedical MoE
TFGformer: Multivariate Time Series Forecasting via Time-Frequency Graph Learning and Covariate FusionRetrieval-augmented generation → Time series and agricultural RAG
QR-Structured Thermal Triggers for Targeted Semantic Attacks on Infrared Vision-Language ModelsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
AgenticRepair: Multi-Faceted Program Context Engineering for Agentic Vulnerability RepairLLM agents and tool use → Software engineering coding agents
Explore Beyond the Boundary Using Entropic InformationDeep reinforcement learning methods → Deep RL algorithms and representations
Role-Break in Attention Heads: Understanding and Detecting Hallucinations in VLMsMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
Beyond Component Testing: Validating Agentic AI SystemsLLM agents and tool use → Agent threats, guardrails and governance
Bridging the Question-Answer Gap in Retrieval-Augmented Generation: Hypothetical Prompt EmbeddingsRetrieval-augmented generation → RAG robustness and adaptive retrieval
OnlineCache: Learning Dynamic Caching Policies with Error Correction for Efficient Diffusion InferenceDiffusion models → Diffusion foundations and sampling
Zero-Mem: Zero-Token Memory Operations for LLM AgentsLLM agents and tool use → Agent memory systems
SAGP: Semantic Affordance-Guided Grasp Planning via Coarse-Zone VLM ReasoningLLM reasoning → Spatial and embodied reasoning
DualDiT: A Conditional Dual-Output Diffusion Transformer for Joint OCT Image and Segmentation Mask GenerationDiffusion models → Medical imaging and segmentation
MAGA: Multi-Platform Self-Fusion of GUI Agents via Structured Action DistillationLLM agents and tool use → GUI and mobile agents
BWM: A Low-Cost High-Fidelity World Simulator for Robot LearningWorld models → Embodied and robot world models
Sample Efficient Hierarchical Reinforcement Learning via Best Policy IdentificationDeep reinforcement learning methods → Deep RL algorithms and representations
Translation with Thought: Difficulty-Adaptive Reasoning via Reinforcement Learning for Multi-Domain Machine TranslationLLM reasoning → Reasoning benchmarks and evaluation
TRACT: Temporally Routed Action Chunks with Chronological Phase Authority for Contact-Rich ManipulationOffline RL and imitation learning from data → Robot manipulation imitation learning

All 411