Papers

New on arXiv

Thu 16 Apr 2026

420 papers

OpenClaw and Ollama in Agentic AI: Toward Fully Autonomous and Scalable AI Agent SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
HG-RAG: Hierarchy-Guided Retrieval-Augmented Generation for Structured Knowledge GraphsRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
Is This Edit Correct? A Multi-Dimensional Benchmark for Reasoning-Aware Image EditingDiffusion models → Image generation, text-to-image and editing
Improving Retrieval-Augmented Generation without Taxonomy-based Error CategorizationRetrieval-augmented generation → Agentic RAG, agents and memory
Task-Level AI Readiness Assessment for Business Process Management:The T-IPO Model and LARA Matrix in Financial-Services IT OperationsLLM agents and tool use → Financial trading and forecasting agents
Reinforcement learning for inverse silhouette design and rapid laser cutting of compact parallelogram quad kirigami prototypesDeep reinforcement learning methods → RL for energy, networks and operations
A pragmatic approach to regulating AI agentsLLM agents and tool use → Agent threats, guardrails and governance
Ave: Guiding Agentic GPU Optimization Using Data-Flow InvariantsLLM agents and tool use → Software engineering coding agents
Understanding Tool-Augmented Agents for Lean Formalization: A Factorial AnalysisLLM agents and tool use → Educational agents and math problem solving
Towards Reliable Testing of Machine UnlearningLLM safety and security → Machine unlearning
SCATR: Simple Calibrated Test-Time RankingLLM reasoning → Test-time compute and efficient thinking
Scaling Test-Time Compute for Agentic CodingLLM reasoning → Reasoning for agents, planning and code +1
CAMP: Cumulative Agentic Masking and Pruning for Privacy Protection in Multi-Turn LLM ConversationsLLM agents and tool use → Agent threats, guardrails and governance
CSLE: A Reinforcement Learning Platform for Autonomous Security ManagementDeep reinforcement learning methods → Adversarial robustness and RL in finance
Don't Make Models Guess Security and Safety: Symbolic Guardrails for Domain-Specific AI AgentsLLM safety and security → Prompt injection and agent security
Reward Weighted Classifier-Free Guidance as Policy Improvement in Autoregressive ModelsLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Why Fine-Tuning Encourages Hallucinations and How to Fix ItKnowledge distillation → Knowledge distillation for LLMs
Subliminal Transfer of Unsafe Behaviors in AI Agent DistillationLLM agents and tool use → Agent threats, guardrails and governance +1
Predicting Where Steering Vectors SucceedMechanistic interpretability → Activation steering methods
Frequency-Aware Flow Matching for High-Quality Image GenerationDiffusion models → Flow matching, time series and science
PolicyBank: Evolving Policy Understanding for LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance
Think Multilingual, Not Harder: A Data-Efficient Framework for Teaching Reasoning Models to Code-SwitchLLM reasoning → Reasoning for agents, planning and code
FineSteer: A Unified Framework for Fine-Grained Inference-Time Steering in Large Language ModelsMechanistic interpretability → Activation steering methods
vstash: Local-First Hybrid Retrieval with Adaptive Fusion for LLM AgentsLLM agents and tool use → Web, search, and research agents
Harmonizing Multi-Objective LLM Unlearning via Unified Domain Representation and Bidirectional Logit DistillationLLM safety and security → Machine unlearning +1
The Semi-Executable Stack: Agentic Software Engineering and the Expanding Scope of SELLM agents and tool use → Software engineering coding agents
Ragged Paged Attention: A High-Performance and Flexible LLM Inference Kernel for TPUEfficient LLM inference and serving → KV cache optimization
RelativeFlow: Taming Medical Image Denoising Learning with Noisy ReferenceDiffusion models → Flow matching, time series and science
DeepER-Med: Advancing Deep Evidence-Based Research in Medicine Through Agentic AILLM agents and tool use → Medical and scientific domain agents
Weak-to-Strong Knowledge Distillation Accelerates Visual LearningKnowledge distillation → General knowledge distillation methods
HarmfulSkillBench: How Do Harmful Skills Weaponize Your Agents?LLM agents and tool use → Agent threats, guardrails and governance
Beyond Single-Model Optimization: Preserving Plasticity in Continual Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
PRL-Bench: A Comprehensive Benchmark Evaluating LLMs' Capabilities in Frontier Physics ResearchLLM agents and tool use → Web, search, and research agents
The Illusion of Equivalence: Systematic FP16 Divergence in KV-Cached Autoregressive InferenceEfficient LLM inference and serving → KV cache optimization
Dispatch-Aware Ragged Attention for Pruned Vision TransformersEfficient LLM inference and serving → Token pruning and merging for ViTs
Analyzing Chain of Thought (CoT) Approaches in Control Flow Code Deobfuscation TasksLLM reasoning → Reasoning for agents, planning and code
LinuxArena: A Control Setting for AI Agents in Live Production Software EnvironmentsLLM agents and tool use → Agent threats, guardrails and governance
LeapAlign: Post-Training Flow Matching Models at Any Generation Step by Building Two-Step TrajectoriesDiffusion models → Flow matching, time series and science
MM-WebAgent: A Hierarchical Multimodal Web Agent for Webpage GenerationLLM agents and tool use → Web, search, and research agents
RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator FrameworkDeep reinforcement learning methods → RL for autonomous driving and traffic
Abstract Sim2Real through Approximate Information StatesDeep reinforcement learning methods → RL for robot locomotion and manipulation
Why Do Vision Language Models Struggle To Recognize Human Emotions?Multimodal LLMs / vision-language models → Multimodal emotion understanding
Enhancing Large Language Models with Retrieval Augmented Generation for Software Testing and Inspection AutomationRetrieval-augmented generation → RAG for code
CoopEval: Benchmarking Cooperation-Sustaining Mechanisms and LLM Agents in Social DilemmasLLM reasoning → Reasoning for agents, planning and code +1
Stability and Generalization in Looped TransformersLLM reasoning → Prompting and chain-of-thought reasoning
From Tokens to Steps: Verification-Aware Speculative Decoding for Efficient Multi-Step ReasoningLLM reasoning → Test-time compute and efficient thinking +1
Agentic Microphysics: A Manifesto for Generative AI SafetyLLM agents and tool use → Agent threats, guardrails and governance
Blue Data Intelligence Layer: Streaming Data and Agents for Multi-source Multi-modal Data-Centric ApplicationsLLM agents and tool use → Web, search, and research agents
RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomographyMultimodal LLMs / vision-language models → Medical vision-language models +1
A Hierarchical Spatiotemporal Action Tokenizer for In-Context Imitation Learning in RoboticsOffline RL and imitation learning from data → Robot manipulation imitation learning
RL-STPA: Adapting System-Theoretic Hazard Analysis for Safety-Critical Reinforcement LearningDeep reinforcement learning methods → Safe and constrained RL
VisPCO: Visual Token Pruning Configuration Optimization via Budget-Aware Pareto-Frontier Learning for Vision-Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Scepsy: Serving Agentic Workflows Using Aggregate LLM PipelinesLLM agents and tool use → Software engineering coding agents
Agent-Aided Design for Dynamic CAD ModelsLLM agents and tool use → Multimodal and embodied agents
MambaSL: Exploring Single-Layer Mamba for Time Series ClassificationState-space sequence models (S4, Mamba) → Time series forecasting with SSMs
An Analysis of Regularization and Fokker-Planck Residuals in Diffusion Models for Image GenerationDiffusion models → Diffusion foundations and sampling
Class Unlearning via Depth-Aware Removal of Forget-Specific DirectionsLLM safety and security → Machine unlearning
Compressing Sequences in the Latent Embedding Space: K-Token Merging for Large Language ModelsEfficient LLM inference and serving → Prompt and context compression
LLMs Gaming Verifiers: RLVR can Lead to Reward HackingLLM reasoning → Reinforcement learning for reasoning +1
IG-Search: Step-Level Information Gain Rewards for Search-Augmented ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning
OpenMobile: Building Open Mobile Agents with Task and Trajectory SynthesisLLM agents and tool use → GUI and mobile agents
Beyond Visual Cues: Semantic-Driven Token Filtering and Expert Routing for Anytime Person ReIDMixture-of-experts → Vision mixture-of-experts
Autonomous Evolution of EDA Tools: Multi-Agent Self-Evolved ABCLLM agents and tool use → General LLM agent frameworks and benchmarks +1
Atropos: Improving Cost-Benefit Trade-off of LLM-based Agents under Self-Consistency with Early Termination and Model HotswapLLM reasoning → Test-time compute and efficient thinking +1
Prefill-as-a-Service: KVCache of Next-Generation Models Could Go Cross-DatacenterEfficient LLM inference and serving → KV cache optimization
From Reactive to Proactive: Assessing the Proactivity of Voice Agents via ProVoice-BenchLLM agents and tool use → General LLM agent frameworks and benchmarks
Autogenesis: A Self-Evolving Agent ProtocolLLM agents and tool use → General LLM agent frameworks and benchmarks +1
Route to Rome Attack: Directing LLM Routers to Expensive Models via Adversarial Suffix OptimizationLLM safety and security → Mixture-of-experts routing safety
DLink: Distilling Layer-wise and Dominant Knowledge from EEG Foundation ModelsKnowledge distillation → Distillation for medical imaging
Towards Faster Language Model Inference Using Mixture-of-Experts Flow MatchingDiffusion models → Discrete and text diffusion +1
Dr. RTL: Autonomous Agentic RTL Optimization through Tool-Grounded Self-ImprovementLLM agents and tool use → Software engineering coding agents +1
Agentic Explainability at Scale: Between Corporate Fears and XAI NeedsLLM agents and tool use → Software engineering coding agents
SAGER: Self-Evolving User Policy Skills for Recommendation AgentLLM agents and tool use → Agentic recommender systems +1
UniDoc-RL: Coarse-to-Fine Visual RAG with Hierarchical Actions and Dense RewardsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
FedGUI: Benchmarking Federated GUI Agents across Heterogeneous Platforms, Devices, and Operating SystemsLLM agents and tool use → GUI and mobile agents
RaTA-Tool: Retrieval-based Tool Selection with Multimodal Large Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
IE as Cache: Information Extraction Enhanced Agentic ReasoningLLM reasoning → Reasoning for agents, planning and code +1
Improving Sparse Autoencoder with Dynamic AttentionMechanistic interpretability → Sparse autoencoder training and evaluation
LongAct: Harnessing Intrinsic Activation Patterns for Long-Context Reinforcement LearningLLM reasoning → Test-time compute and efficient thinking +1
Dual-Axis Generative Reward Model Toward Semantic and Turn-taking Robustness in Interactive Spoken Dialogue ModelsLLM alignment / RLHF / preference optimisation → Preference alignment for recommenders and RAG
Reward-Aware Trajectory Shaping for Few-step Visual GenerationLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
Governing Reflective Human-AI Collaboration: A Framework for Epistemic Scaffolding and Traceable ReasoningLLM reasoning → Reasoning benchmarks and evaluation
Toward Agentic RAG for UkrainianRetrieval-augmented generation → Agentic RAG, agents and memory +1
Beyond Importance Sampling: Rejection-Gated Policy OptimizationLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
Can LLMs Accurately Score Medical Diagnoses and Clinical Reasoning?LLM reasoning → Clinical and biomedical applications
MemoSight: Unifying Context Compression and Multi Token Prediction for Reasoning AccelerationLLM reasoning → Test-time compute and efficient thinking +1
Reasoning Dynamics and the Limits of Monitoring Modality Reliance in Vision-Language ModelsLLM reasoning → Multimodal reasoning +1
RACER: Retrieval-Augmented Contextual Rapid Speculative DecodingRetrieval-augmented generation → Generation and context integration +1
The Missing Knowledge Layer in AI: A Framework for Stable Human-AI ReasoningLLM reasoning → Reasoning benchmarks and evaluation
Does RL Expand the Capability Boundary of LLM Agents? A PASS@(k,T) AnalysisLLM agents and tool use → Reinforcement learning for LLM agents
SkillDroid: Compile Once, Reuse ForeverLLM agents and tool use → Agent skill learning and evolution
MetaDent: Labeling Clinical Images for Vision-Language Models in DentistryMultimodal LLMs / vision-language models → Medical vision-language models
Segment-Level Coherence for Robust Harmful Intent Probing in LLMsLLM safety and security → Jailbreaking
Adaptive Test-Time Compute Allocation for Reasoning LLMs via Constrained Policy OptimizationLLM reasoning → Test-time compute and efficient thinking
TrigReason: Trigger-Based Collaboration between Small and Large Reasoning ModelsLLM reasoning → Test-time compute and efficient thinking
Switch: Learning Agile Skills Switching for Humanoid RobotsDeep reinforcement learning methods → RL for robot locomotion and manipulation
SWE-TRACE: Optimizing Long-Horizon SWE Agents Through Rubric Process Reward Models and Heuristic Test-Time ScalingLLM reasoning → Reasoning for agents, planning and code +1
Learning Ad Hoc Network Dynamics via Graph-Structured World ModelsWorld models → World models for language agents
Modeling LLM Unlearning as an Asymmetric Two-Task Learning ProblemLLM safety and security → Machine unlearning
Knowing When Not to Answer: Evaluating Abstention in Multimodal Reasoning SystemsLLM reasoning → Multimodal reasoning

All 420