Papers

New on arXiv

Mon 27 Apr 2026

425 papers

Less is More: Lightweight Prompt Compression for Question Answering Applications on Edge DevicesRetrieval-augmented generation → Generation and context integration +1
Detecting Hallucinations for Large Language Model-based Knowledge Graph ReasoningLLM reasoning → Hallucination detection and mitigation
Pruning via Causal Attribution Preserves Reasoning Performance in Large Language ModelsLLM reasoning → Test-time compute and efficient thinking
Thoughts-as-Planning: Latent World Models for Chain-of-Thoughts Optimization via Reinforcement PlanningWorld models → World models for language agents +1
Statistical Inference and Quality Measures of KV Cache Quantisations Inspired by TurboQuantEfficient LLM inference and serving → KV cache quantization
Context-Augmented Code Generation: How Product Context Improves AI Coding Agent Decision Compliance by 49%LLM agents and tool use → Software engineering coding agents
X2SAM: Any Segmentation in Images and VideosMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Selective Correlation Based Knowledge Distillation for Ground Reaction Force EstimationKnowledge distillation → Distillation for medical imaging
LiteVLA-H: Dual-Rate Vision-Language-Action Inference for Onboard Aerial Guidance and Semantic PerceptionVision-language-action models → Aerial and drone VLA
Towards High Fidelity Face Swapping: A Comprehensive Survey and New BenchmarkDiffusion models → Image generation, text-to-image and editing
Frontier Coding Agents Can Now Implement an AlphaZero Self-Play Machine Learning Pipeline For Connect Four That Performs Comparably to an External SolverLLM agents and tool use → Web, search, and research agents +1
Learning biophysical models of gene regulation with probability flow matchingDiffusion models → Flow matching, time series and science
Analyzing LLM Reasoning to Uncover Mental Health StigmaLLM reasoning → Clinical and biomedical applications
DiscreteRTC: Discrete Diffusion Policies are Natural Asynchronous ExecutorsDiffusion models → Diffusion and flow policies for robotics
Dual-Track CoT: Budget-Aware Stepwise Guidance for Small LMsLLM reasoning → Test-time compute and efficient thinking
Why Search When You Can Transfer? Amortized Agentic Workflow Design from Structural PriorsLLM agents and tool use → Software engineering coding agents
Why Does Reinforcement Learning Generalize? A Feature-Level Mechanistic Study of Post-Training in Large Language ModelsLLM reasoning → Reinforcement learning for reasoning
Sparse Personalized Text Generation with Multi-Trajectory ReasoningLLM reasoning → Reinforcement learning for reasoning
Adaptive Prompt Embedding Optimization for LLM JailbreakingLLM safety and security → Jailbreaking
PolyKV: A Shared Asymmetrically-Compressed KV Cache Pool for Multi-Agent LLM InferenceLLM agents and tool use → Software engineering coding agents +1
Odysseys: Benchmarking Web Agents on Realistic Long Horizon TasksLLM agents and tool use → Web, search, and research agents
Compute Aligned Training: Optimizing for Test Time InferenceLLM reasoning → Test-time compute and efficient thinking
BenchGuard: Who Guards the Benchmarks? Automated Auditing of LLM Agent BenchmarksLLM agents and tool use → General LLM agent frameworks and benchmarks
ViPO: Visual Preference Optimization at ScaleLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
Learning from Noisy Preferences: A Semi-Supervised Learning Approach to Direct Preference OptimizationLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
Rethinking Layer Redundancy: Calibration Matters More Than Search in LLM Depth PruningLLM reasoning → Test-time compute and efficient thinking
A Unifying Framework for Unsupervised Concept ExtractionMechanistic interpretability → Sparse autoencoder training and evaluation
CAN-QA: A Question-Answering Benchmark for Reasoning over In-Vehicle CAN TrafficLLM reasoning → Chain-of-thought reasoning for driving
S-SONDO: Self-Supervised Knowledge Distillation for General Audio Foundation ModelsKnowledge distillation → Speech and audio distillation
Large Language Models Explore by Latent DistillingLLM reasoning → Test-time compute and efficient thinking
Libra-VLA: Achieving Learning Equilibrium via Asynchronous Coarse-to-Fine Dual-SystemVision-language-action models → Efficient VLA inference and adaptation
SUDP: Secret-Use Delegation Protocol for Agentic SystemsLLM agents and tool use → Agent threats, guardrails and governance
Agentic AI for Remote Sensing: Technical Challenges and Research DirectionsLLM agents and tool use → Multimodal and embodied agents
asRoBallet: Closing the Sim2Real Gap via Friction-Aware Reinforcement Learning for Underactuated Spherical DynamicsDeep reinforcement learning methods → RL for robot locomotion and manipulation
Generative diffusion models for spatiotemporal influenza forecastingDiffusion models → Flow matching, time series and science
MultiHedge: Adaptive Coordination via Retrieval-Augmented ControlRetrieval-augmented generation → Agentic RAG, agents and memory
VibeToken: Scaling 1D Image Tokenizers and Autoregressive Models for Dynamic Resolution GenerationsDiffusion models → Diffusion foundations and sampling
Latent Agents: A Post-Training Procedure for Internalized Multi-Agent DebateLLM reasoning → Reasoning for agents, planning and code +1
Learning Illumination Control in Diffusion ModelsDiffusion models → 3D, point cloud and scene generation
Co-Director: Agentic Generative Video StorytellingDiffusion models → Video generation +1
On the Trainability of Masked Diffusion Language Models via Blockwise LocalityDiffusion models → Discrete and text diffusion
FGDM: Reasoning Aware Multi-Agentic Framework for Software Bug Detection using Chain of Thought and Tree of Thought PromptingLLM reasoning → Reasoning for agents, planning and code +1
A Comparative Evaluation of AI Agent Security GuardrailsLLM safety and security → Prompt injection and agent security
Programming with Data: Test-Driven Data Engineering for Self-Improving LLMs from Raw CorporaSynthetic data and self-improvement for LLMs → Code and instruction data synthesis
Nautile-370M: Spectral Memory Meets Attention in a Small Reasoning ModelLLM reasoning → Test-time compute and efficient thinking
ITAS: A Multi-Agent Architecture for LLM-Based Intelligent TutoringLLM agents and tool use → Educational agents and math problem solving
Intrinsic Mutual Information as a Modulator for Preference OptimizationLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Learning to Think from Multiple ThinkersLLM reasoning → Prompting and chain-of-thought reasoning
SpecRLBench: A Benchmark for Generalization in Specification-Guided Reinforcement LearningDeep reinforcement learning methods → Interpretable and logic-guided RL
Green Shielding: A User-Centric Approach Towards Trustworthy AILLM safety and security → Safety alignment, refusal and guardrails
Can Current Agents Close the Discovery-to-Application Gap? A Case Study in MinecraftLLM agents and tool use → Web, search, and research agents
NeuroClaw Technical ReportLLM agents and tool use → Medical and scientific domain agents
Contextual Linear Activation Steering of Language ModelsMechanistic interpretability → Activation steering methods
The Price of Agreement: Measuring LLM Sycophancy in Agentic Financial ApplicationsLLM agents and tool use → Financial trading and forecasting agents
Benchmarking Source-Sensitive Reasoning in Turkish: Humans and LLMs under Evidential Trust ManipulationLLM reasoning → Reasoning benchmarks and evaluation
AgentWard: A Lifecycle Security Architecture for Autonomous AI AgentsLLM agents and tool use → Agent threats, guardrails and governance
DepthKV: Layer-Dependent KV Cache Pruning for Long-Context LLM InferenceEfficient LLM inference and serving → KV cache optimization
K-MetBench: A Multi-Dimensional Benchmark for Fine-Grained Evaluation of Expert Reasoning, Locality, and Multimodality in MeteorologyLLM reasoning → Time series reasoning
Cortex-Inspired Continual Learning: Unsupervised Instantiation and Recovery of Functional Task NetworksMixture-of-experts → Mixture of LoRA experts
Meta-CoT: Enhancing Granularity and Generalization in Image EditingLLM reasoning → Multimodal reasoning
XGRAG: A Graph-Native Framework for Explaining KG-based Retrieval-Augmented GenerationRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
CF-VLA: Efficient Coarse-to-Fine Action Generation for Vision-Language-Action PoliciesVision-language-action models → Efficient VLA inference and adaptation
Evaluating whether AI models would sabotage AI safety researchLLM agents and tool use → Agent threats, guardrails and governance
Majorization-Guided Test-Time Adaptation for Vision-Language Models under Modality-Specific ShiftMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Skill Retrieval Augmentation for Agentic AILLM agents and tool use → Agent skill learning and evolution
A systematic evaluation of vision-language models for observational astronomical reasoning tasksLLM reasoning → Knowledge- and tool-augmented reasoning +1
Improving Vision-language Models with Perception-centric Process Reward ModelsLLM reasoning → Multimodal reasoning +2
Measuring the Unmeasurable: Markov Chain Reliability for LLM AgentsLLM agents and tool use → Software engineering coding agents
Diffusion Model as a Generalist Segmentation LearnerDiffusion models → Medical imaging and segmentation
FastOMOP: A Foundational Architecture for Reliable Agentic Real-World Evidence Generation on OMOP CDM dataLLM agents and tool use → Medical and scientific domain agents
MEG-RAG: Quantifying Multi-modal Evidence Grounding for Evidence Selection in RAGMultimodal LLMs / vision-language models → Document understanding and multimodal RAG +1
Towards Lawful Autonomous Driving: Deriving Scenario-Aware Driving Requirements from Traffic Laws and RegulationsEnd-to-end autonomous driving → Large language models for driving
Hierarchical Behaviour SpacesDeep reinforcement learning methods → Deep RL algorithms and representations
Layerwise Convergence Fingerprints for Runtime Misbehavior Detection in Large Language ModelsLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
A Reward-Free Viewpoint on Multi-Objective Reinforcement LearningDeep reinforcement learning methods → RL for games and combinatorial optimization
SEARCH-R: Structured Entity-Aware Retrieval with Chain-of-Reasoning Navigator for Multi-hop Question AnsweringLLM reasoning → Knowledge- and tool-augmented reasoning
Beyond the Attention Stability Boundary: Agentic Self-Synthesizing Reasoning ProtocolsLLM reasoning → Reasoning for agents, planning and code +1
Guiding Vector Field Generation via Score-based Diffusion ModelDiffusion models → Diffusion and flow policies for robotics
Zero-to-CAD: Agentic Synthesis of Interpretable CAD Programs at Million-Scale Without Real DataLLM agents and tool use → Multimodal and embodied agents
GAMMAF: A Common Framework for Graph-Based Anomaly Monitoring Benchmarking in LLM Multi-Agent SystemsLLM agents and tool use → Agent threats, guardrails and governance
Agentic clinical reasoning over longitudinal myeloma records: a retrospective evaluation against expert consensusLLM reasoning → Clinical and biomedical applications
On the Footprints of Reviewer Bots Feedback on Agentic Pull Requests in OSS GitHub RepositoriesLLM agents and tool use → Software engineering coding agents
Characterizing Vision-Language-Action Models across XPUs: Constraints and Acceleration for On-Robot DeploymentVision-language-action models → Efficient VLA inference and adaptation
PhysNote: Self-Knowledge Notes for Evolvable Physical Reasoning in Vision-Language ModelLLM reasoning → Spatial and embodied reasoning +1
Kwai Summary Attention Technical ReportLLM reasoning → Test-time compute and efficient thinking +2
A Multi-Dimensional Audit of Politically Aligned Large Language ModelsLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
AD-Relight: Training-Free Banner Relighting via Illumination Translation with Diffusion PriorsDiffusion models → 3D, point cloud and scene generation
How Personal Characteristics Shape User Exploration of Diverse Movie Recommendations with a LLM-Based Multi-Agent SystemLLM agents and tool use → Agentic recommender systems
An Automatic Ground Collision Avoidance System with Reinforcement LearningDeep reinforcement learning methods → RL for robot navigation and UAV control
All That Glitters Is Not Audio: Rethinking Text Priors and Audio Reliance in Audio-Language EvaluationLLM reasoning → Audio-language model reasoning
Global Context or Local Detail? Adaptive Visual Grounding for Hallucination MitigationMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
Aligning with Your Own Voice: Self-Corrected Preference Learning for Hallucination Mitigation in LVLMsMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs +1
Complexity of Linear Regions in Self-supervised Deep ReLU NetworksKnowledge distillation → General knowledge distillation methods
Structural Pruning of Large Vision Language Models: A Comprehensive Study on Pruning Dynamics, Recovery, and Data EfficiencyMultimodal LLMs / vision-language models → Visual token pruning and efficiency
DPRM: A Plug-in Doob h transform-induced Token-Ordering Module for Discrete Diffusion ModelsDiffusion models → Discrete and text diffusion
An Aircraft Upset Recovery System with Reinforcement LearningDeep reinforcement learning methods → RL for robot navigation and UAV control
Diffusion Templates: A Unified Plugin Framework for Controllable DiffusionDiffusion models → Diffusion foundations and sampling
OS-SPEAR: A Toolkit for the Safety, Performance,Efficiency, and Robustness Analysis of OS AgentsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
SycoPhantasy: Quantifying Sycophancy and Hallucination in Small Open Weight VLMs for Vision-Language Scoring of Fantasy CharactersMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
See Further, Think Deeper: Advancing VLM's Reasoning Ability with Low-level Visual Cues and ReflectionLLM reasoning → Multimodal reasoning +1

All 425