Papers

New on arXiv

Tue 5 May 2026

430 papers

SDAD: Spec-Driven Agentic Development for the AI-Native SDLCLLM agents and tool use → Software engineering coding agents
Concept-based Visual Counterfactual Explanations with Diffusion ModelsDiffusion models → Diffusion foundations and sampling
A Unified Approach to Interpreting Knowledge Distillation for Large Language Models via InteractionsKnowledge distillation → Knowledge distillation for LLMs
Performance Optimization and Comparative Analysis of Generative AI Models on Advanced AcceleratorsDiffusion models → Diffusion foundations and sampling
Tile-Level Activation Overlap for Efficient LLM InferenceEfficient LLM inference and serving → Serving systems and request scheduling
An Ethical eValuation Agent (EeVA): Results of a Proof-of-Concept Test on a Prototype Agentic-like Workflow to Assist Ethical DeliberationsLLM agents and tool use → General LLM agent frameworks and benchmarks
Evaluating Advanced Prompting on Gemini Flash for Multi-Hop Biomedical QALLM reasoning → Clinical and biomedical applications
Weakly Supervised Concept Learning for Object-centric Visual ReasoningLLM reasoning → Multimodal reasoning
Where Reliability Lives in Vision-Language Models: A Mechanistic Study of Attention, Hidden States, and Causal CircuitsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Rethinking Entropy Minimization in Test-Time Adaptation for Autoregressive ModelsDeep reinforcement learning methods → RL training of language and agent models
Quantile Geometry Regularization for Distributional Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
KARMA-MV: A Benchmark for Causal Question Answering on Music VideosLLM reasoning → Audio-language model reasoning
Multimodal synthesis of MRI and tabular data with diffusion in a joint latent space via cross-attentionDiffusion models → Medical imaging and segmentation
EnterpriseRAG-Bench: A RAG Benchmark for Company Internal KnowledgeRetrieval-augmented generation → RAG pipeline design and evaluation
Decision-aware User Simulation Agent for Evaluating Conversational Recommender SystemsLLM agents and tool use → Social simulation and multi-agent debate
Coral: Cost-Efficient Multi-LLM Serving over Heterogeneous Cloud GPUsEfficient LLM inference and serving → Serving systems and request scheduling
Efficiently Aligning Language Models with Online Natural Language FeedbackLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Probing Structural Mathematical Reasoning in Language Models with Algebraic TrapdoorsLLM reasoning → Mathematical reasoning
The Scaling Properties of Implicit Deductive Reasoning in TransformersLLM reasoning → Prompting and chain-of-thought reasoning
NoisyCausal: A Benchmark for Evaluating Causal Reasoning Under Structured NoiseLLM reasoning → Causal reasoning in LLMs
Agent Island: A Saturation- and Contamination-Resistant Benchmark from Multiagent GamesLLM agents and tool use → Reinforcement learning for LLM agents
Hierarchical Visual Agent: Managing Contexts in Joint Image-Text Space for Advanced Chart ReasoningLLM reasoning → Multimodal reasoning +1
Leveraging Pretrained Language Models as Energy Functions for Glauber Dynamics Text DiffusionDiffusion models → Discrete and text diffusion
Material Database Agent: A Multimodal Agentic Framework for Scientific Literature MiningLLM agents and tool use → Medical and scientific domain agents
Explaining and Preventing Alignment Collapse in Iterative RLHFLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Governed Collaborative Memory as Artificial Selection in LLM-Based Multi-Agent SystemsLLM agents and tool use → Agent memory systems
Parallel Prefix Verification for Speculative GenerationEfficient LLM inference and serving → Speculative decoding
Laundering AI Authority with Adversarial ExamplesMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Hierarchical Support Vector State Partitioning for Distilling Black Box Reinforcement Learning PoliciesDeep reinforcement learning methods → Deep RL algorithms and representations
Temporal Reasoning Is Not the Bottleneck: A Probabilistic Inconsistency Framework for Neuro-Symbolic QALLM reasoning → Reasoning benchmarks and evaluation
Adaptive Consensus in LLM Ensembles via Sequential Evidence Accumulation: Automatic Budget Identification and Calibrated Commit SignalsLLM reasoning → Test-time compute and efficient thinking
Constraint-Enhanced Reinforcement Learning Based on Dynamic Decoupled Spherical Radial SquashingDeep reinforcement learning methods → RL for robot locomotion and manipulation
Position: the Stochastic Parrot in the Coal Mine. Model Collapse is a Threat to Low-Resource CommunitiesSynthetic data and self-improvement for LLMs → Model collapse from synthetic data
Membership Inference Attacks for Retrieval Based In-Context Learning for Document Question AnsweringLLM safety and security → Membership inference attacks
Large Language Models are Universal Reasoners for Visual GenerationLLM reasoning → Multimodal reasoning +1
Safety and accuracy follow different scaling laws in clinical large language modelsLLM reasoning → Clinical and biomedical applications +2
OpenSeeker-v2: Pushing the Limits of Search Agents with Informative and High-Difficulty TrajectoriesLLM agents and tool use → Web, search, and research agents
Stayin' Aligned Over Time: Towards Longitudinal Human-LLM Alignment via Contextual Reflection and Privacy-Preserving Behavioral DataLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Redefining AI Red Teaming in the Agentic Era: From Weeks to HoursLLM agents and tool use → Agent threats, guardrails and governance +1
Rethinking Reasoning-Intensive Retrieval: Evaluating and Advancing Retrievers in Agentic Search SystemsLLM agents and tool use → Web, search, and research agents
Conditional Diffusion SamplingDiffusion models → Diffusion foundations and sampling
SymptomAI: Toward a Conversational AI Agent for Everyday Symptom AssessmentLLM agents and tool use → Medical and scientific domain agents
Physics-Grounded Multi-Agent Architecture for Traceable, Risk-Aware Human-AI Decision Support in ManufacturingLLM agents and tool use → Multimodal and embodied agents
RD-ViT: Recurrent-Depth Vision Transformer for Semantic Segmentation with Reduced Data Dependence Extending the Recurrent-Depth Transformer Architecture to Dense PredictionMixture-of-experts → Vision mixture-of-experts
An Agent-Oriented Pluggable Experience-RAG Skill for Experience-Driven Retrieval Strategy OrchestrationLLM reasoning → Knowledge- and tool-augmented reasoning +1
From Intent to Execution: Composing Agentic Workflows with Agent RecommendationLLM agents and tool use → Agentic recommender systems
Flow Sampling: Learning to Sample from Unnormalized Densities via Denoising Conditional ProcessesDiffusion models → Flow matching, time series and science
MOSAIC-Bench: Measuring Compositional Vulnerability Induction in Coding AgentsLLM agents and tool use → Agent threats, guardrails and governance
UnAC: Adaptive Visual Prompting with Abstraction and Stepwise Checking for Complex Multimodal ReasoningLLM reasoning → Multimodal reasoning
iWorld-Bench: A Benchmark for Interactive World Models with a Unified Action Generation FrameworkWorld models → Video-generation world models and simulators
StateVLM: A State-Aware Vision-Language Model for Robotic Affordance ReasoningLLM reasoning → Spatial and embodied reasoning +1
Steer Like the LLM: Activation Steering that Mimics PromptingMechanistic interpretability → Activation steering methods
CC-OCR V2: Fine-Grained Attribution of LMM Failures in Real-World Visual Document UnderstandingMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Contextual Multi-Objective Optimization: Rethinking Objectives in Frontier AI SystemsLLM reasoning → Reasoning for agents, planning and code
QKVShare: Quantized KV-Cache Handoff for Multi-Agent On-Device LLMsLLM agents and tool use → Software engineering coding agents +1
DMGD: Train-Free Dataset Distillation with Semantic-Distribution Matching in Diffusion ModelsDiffusion models → Diffusion foundations and sampling
EvoLM: Self-Evolving Language Models through Co-Evolved Discriminative RubricsLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
Quantifying the human visual exposome with vision language modelsMultimodal LLMs / vision-language models → Multimodal emotion understanding
Correct Is Not Enough: Training Reasoning Planners with Executor-Grounded RewardsLLM reasoning → Reinforcement learning for reasoning +1
Evaluating Generative Models as Interactive Emergent Representations of Human-Like Collaborative BehaviorLLM agents and tool use → General LLM agent frameworks and benchmarks
Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video GenerationDiffusion models → Video generation +1
TRACE: A Metrologically-Grounded Engineering Framework for Trustworthy Agentic AI Systems in Operationally Critical DomainsLLM agents and tool use → Agent threats, guardrails and governance
RoboAlign-R1: Distilled Multimodal Reward Alignment for Robot Video World ModelsWorld models → Embodied and robot world models +1
Agentic-imodels: Evolving agentic interpretability tools via autoresearchLLM agents and tool use → Web, search, and research agents
ScrapMem: A Bio-inspired Framework for On-device Personalized Agent Memory via Optical ForgettingLLM agents and tool use → Agent memory systems
Towards accurate extreme event likelihoods from diffusion model climate emulatorsDiffusion models → Flow matching, time series and science
Natural Language Processing: A Comprehensive Practical Guide from Tokenisation to RLHFRetrieval-augmented generation → Generation and context integration +1
Enhancing Visual Question Answering with Multimodal LLMs via Chain-of-Question Guided Retrieval-Augmented GenerationMultimodal LLMs / vision-language models → Document understanding and multimodal RAG +1
Say the Mission, Execute the Swarm: Agent-Enhanced LLM Reasoning in the Web-of-DronesLLM reasoning → Spatial and embodied reasoning +1
What You Think is What You See: Driving Exploration in VLM Agents via Visual-Linguistic CuriosityWorld models → World models for language agents +1
GeoTopoDiff: Learning Geometry--Topology Graph Priors through Boundary-Constrained Mixed Diffusion for Sparse-Slice 3D Porous ReconstructionDiffusion models → Flow matching, time series and science
Before Forgetting, Learn to Remember: Revisiting Foundational Learning Failures in LVLM Unlearning BenchmarksMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Vanishing L2 regularization for the softmax Multi Armed BanditDeep reinforcement learning methods → RL theory and sample complexity
Rose-SQL: Role-State Evolution Guided Structured Reasoning for Multi-Turn Text-to-SQLLLM reasoning → Knowledge- and tool-augmented reasoning
Unified Multimodal Visual Tracking with Dual Mixture-of-ExpertsMixture-of-experts → Vision mixture-of-experts
Tempered Guided DiffusionDiffusion models → Diffusion foundations and sampling
Uni-OPD: Unifying On-Policy Distillation with a Dual-Perspective RecipeKnowledge distillation → On-policy and self-distillation for LLMs
MEMTIER: Tiered Memory Architecture and Retrieval Bottleneck Analysis for Long-Running Autonomous AI AgentsLLM agents and tool use → Agent memory systems
AdapShot: Adaptive Many-Shot In-Context Learning with Semantic-Aware KV Cache ReuseLLM reasoning → Test-time compute and efficient thinking +1
The Detector Teaches Itself: Lightweight Self-Supervised Adaptation for Open-Vocabulary Object DetectionMultimodal LLMs / vision-language models → Driving VLMs and open-vocabulary detection
Diffusion Masked Pretraining for Dynamic Point CloudDiffusion models → 3D, point cloud and scene generation
Where Paths Split: Localized, Calibrated Control of Moral Reasoning in Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
Multi-Agent Strategic Games with LLMsLLM agents and tool use → Social simulation and multi-agent debate
Workspace-Bench 1.0: Benchmarking AI Agents on Workspace Tasks with Large-Scale File DependenciesLLM agents and tool use → General LLM agent frameworks and benchmarks
HeadQ: Model-Visible Distortion and Score-Space Correction for KV-Cache QuantizationEfficient LLM inference and serving → KV cache quantization
PerFlow: Physics-Embedded Rectified Flow for Efficient Reconstruction and Uncertainty Quantification of Spatiotemporal DynamicsDiffusion models → Flow matching, time series and science
Erase Persona, Forget Lore: Benchmarking Multimodal Copyright Unlearning in Large Vision Language ModelsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
A Skill-Based AI Agentic Pipeline for Library of Congress Subject IndexingLLM agents and tool use → Web, search, and research agents
SURE-RAG: Sufficiency and Uncertainty-Aware Evidence Verification for Selective Retrieval-Augmented GenerationRetrieval-augmented generation → RAG robustness and adaptive retrieval
LATS-RCA: Language Agent Tree Search for Root Cause Analysis in MicroservicesLLM agents and tool use → Software engineering coding agents
GRIFDIR: Graph Resolution-Invariant FEM Diffusion Models in Function Spaces over Irregular DomainsDiffusion models → Diffusion foundations and sampling
Real-Time Evaluation of Autonomous Systems under Adversarial AttacksEnd-to-end autonomous driving → Adversarial scenarios and safety testing +1
MHPR: Multidimensional Human Perception and Reasoning Benchmark for Large Vision-Languate ModelsLLM reasoning → Multimodal reasoning
MEMSAD: Gradient-Coupled Anomaly Detection for Memory Poisoning in Retrieval-Augmented AgentsRetrieval-augmented generation → Privacy and security of RAG +1
CuraView: A Multi-Agent Framework for Medical Hallucination Detection with GraphRAG-Enhanced Knowledge VerificationLLM agents and tool use → Medical and scientific domain agents
FinSTaR: Towards Financial Reasoning with Time Series Reasoning ModelsLLM reasoning → Reasoning for agents, planning and code
Exposing LLM Safety Gaps Through Mathematical Encoding:New Attacks and Systematic AnalysisLLM safety and security → Safety alignment, refusal and guardrails
Mantis: Mamba-native Tuning is Efficient for 3D Point Cloud Foundation ModelsState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
Quantum Hierarchical Reinforcement Learning via Variational Quantum CircuitsDeep reinforcement learning methods → Quantum reinforcement learning
Replacing Parameters with Preferences: Federated Alignment of Heterogeneous Vision-Language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP +1

All 430