Papers

New on arXiv

Mon 20 Oct 2025

404 papers

KP-Agent: Keyword Pruning in Sponsored Search Advertising via LLM-Powered Contextual BanditsLLM agents and tool use → Web, search, and research agents
Naiad: Novel Agentic Intelligent Autonomous System for Inland Water MonitoringLLM agents and tool use → Multimodal and embodied agents
DynaKV: Enabling Accurate and Efficient Long-Sequence LLM Decoding on SmartphonesEfficient LLM inference and serving → KV cache optimization
Network and Systems Performance Characterization of MCP-Enabled LLM AgentsLLM agents and tool use → Tool learning and function calling
It Takes Two to Tango: Two Parallel Samplers Improve Quality in Diffusion Models for Limited StepsDiffusion models → Diffusion foundations and sampling
Token-Level Inference-Time Alignment for Vision-Language ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
Variance-Reduction Guidance: Sampling Trajectory Optimization for Diffusion ModelsDiffusion models → Diffusion foundations and sampling
TrustResearcher: Automating Knowledge-Grounded and Transparent Research Ideation with Multi-Agent CollaborationLLM agents and tool use → Web, search, and research agents
CourtGuard: A Local, Multiagent Prompt Injection ClassifierLLM safety and security → Prompt injection and agent security
OmniCast: A Masked Latent Diffusion Model for Weather Forecasting Across Time ScalesDiffusion models → Flow matching, time series and science
Local Coherence or Global Validity? Investigating RLVR Traces in Math DomainsLLM reasoning → Reinforcement learning for reasoning
Saber: An Efficient Sampling with Adaptive Acceleration and Backtracking Enhanced Remasking for Diffusion Language ModelDiffusion models → Discrete and text diffusion
LLM-Based Multi-Agent System for Simulating and Analyzing Marketing and Consumer BehaviorLLM agents and tool use → Social simulation and multi-agent debate
Annotating the Chain-of-Thought: A Behavior-Labeled Dataset for AI SafetyLLM reasoning → Reasoning for agents, planning and code +2
Extracting Rule-based Descriptions of Attention Features in TransformersMechanistic interpretability → Circuits and mechanistic analysis
World-in-World: World Models in a Closed-Loop WorldWorld models → World models for language agents
Evaluating LLM Reasoning Beyond Correctness and CoTLLM reasoning → Reasoning benchmarks and evaluation
BlueCodeAgent: A Blue Teaming Agent Enabled by Automated Red Teaming for CodeGen AILLM agents and tool use → Agent threats, guardrails and governance +1
SafeCoop: Unravelling Full Stack Safety in Agentic Collaborative DrivingEnd-to-end autonomous driving → Cooperative and V2X driving +1
HyperDiffusionFields (HyDiF): Diffusion-Guided Hypernetworks for Learning Implicit Molecular Neural FieldsDiffusion models → Flow matching, time series and science
Gradient Variance Reveals Failure Modes in Flow-Based Generative ModelsDiffusion models → Flow matching, time series and science
Online In-Context Distillation for Low-Resource Vision Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Latent-Augmented Discrete Diffusion ModelsDiffusion models → Discrete and text diffusion
Investigating the Impact of Dark Patterns on LLM-Based Web AgentsLLM agents and tool use → Agent threats, guardrails and governance
Does Reasoning Help LLM Agents Play Dungeons and Dragons? A Prompt Engineering ExperimentLLM reasoning → Reasoning for agents, planning and code +1
SMaRT: Select, Mix, and ReinvenT -- A Strategy Fusion Framework for LLM-Driven Reasoning and PlanningLLM reasoning → Reasoning for agents, planning and code
Planned DiffusionDiffusion models → Discrete and text diffusion
R2BC: Multi-Agent Imitation Learning from Single-Agent DemonstrationsOffline RL and imitation learning from data → Imitation learning methods
Provably Optimal Reinforcement Learning under Safety FilteringDeep reinforcement learning methods → Safe and constrained RL
Any-Depth Alignment: Unlocking Innate Safety Alignment of LLMs to Any-DepthLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
Chain-of-Thought Reasoning Improves Context-Aware Translation with Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
R2L: Reliable Reinforcement Learning: Guaranteed Return & Reliable Policies in Reinforcement LearningDeep reinforcement learning methods → Safe and constrained RL
Fine-tuning Flow Matching Generative Models with Intermediate FeedbackDiffusion models → Diffusion and flow policies for robotics +1
Adaptive Divergence Regularized Policy Optimization for Fine-tuning Generative ModelsLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
CompactPrompt: A Unified Pipeline for Prompt Data Compression in LLM WorkflowsLLM reasoning → Test-time compute and efficient thinking +1
OPTAGENT: Optimizing Multi-Agent LLM Interactions Through Verbal Reinforcement Learning for Enhanced ReasoningLLM reasoning → Reasoning for agents, planning and code +1
Is Multilingual LLM Watermarking Truly Multilingual? Scaling Robustness to 100+ Languages via Back-TranslationLLM safety and security → Watermarking generated content and models
BadScientist: Can a Research Agent Write Convincing but Unsound Papers that Fool LLM Reviewers?LLM agents and tool use → Web, search, and research agents
Humanoid Goalkeeper: Learning from Position Conditioned Task-Motion ConstraintsDeep reinforcement learning methods → RL for robot locomotion and manipulation
FABRIC: Framework for Agent-Based Realistic Intelligence CreationLLM agents and tool use → General LLM agent frameworks and benchmarks
Demystifying Transition Matching: When and Why It Can Beat Flow MatchingDiffusion models → Flow matching, time series and science
PLAGUE: Plug-and-play framework for Lifelong Adaptive Generation of Multi-turn ExploitsLLM safety and security → Jailbreaking
UniRL-Zero: Reinforcement Learning on Unified Models with Joint Language Model and Diffusion Model ExpertsLLM reasoning → Reinforcement learning for reasoning +2
AtlasKV: Augmenting LLMs with Billion-Scale Knowledge Graphs in 20GB VRAMRetrieval-augmented generation → RAG pipeline design and evaluation
From Charts to Code: A Hierarchical Benchmark for Multimodal ModelsMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
EvoSyn: Generalizable Evolutionary Data Synthesis for Verifiable LearningLLM agents and tool use → General LLM agent frameworks and benchmarks +1
SpecAgent: A Speculative Retrieval and Forecasting Agent for Code CompletionLLM agents and tool use → Software engineering coding agents
Rewarding the Journey, Not Just the Destination: A Composite Path and Answer Self-Scoring Reward Mechanism for Test-Time Reinforcement LearningLLM alignment / RLHF / preference optimisation → Process reward models for reasoning +1
Select-Then-Decompose: From Empirical Analysis to Adaptive Selection Strategy for Task Decomposition in Large Language ModelsLLM reasoning → Test-time compute and efficient thinking
CLAWS:Creativity detection for LLM-generated solutions using Attention Window of SectionsLLM reasoning → Reasoning benchmarks and evaluation
ParaVul: A Parallel Large Language Model and Retrieval-Augmented Framework for Smart Contract Vulnerability DetectionRetrieval-augmented generation → Privacy and security of RAG
Not Every Time and Frequency Need to Be Forgotten in Diffusion UnlearningDiffusion models → Robustness, privacy and adversarial aspects +1
Robobench: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models as Embodied BrainMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
SoftMimic: Learning Compliant Whole-body Control from ExamplesDeep reinforcement learning methods → RL for robot locomotion and manipulation
UltraCUA: A Foundation Model for Computer Use Agents with Hybrid ActionLLM agents and tool use → GUI and mobile agents
Inference-Time Compute Scaling For Flow MatchingLLM reasoning → Diffusion reasoning and test-time scaling +1
SparseVILA: Decoupling Visual Sparsity for Efficient VLM InferenceMultimodal LLMs / vision-language models → Visual token pruning and efficiency
Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
VERA-V: Variational Inference Framework for Jailbreaking Vision-Language ModelsLLM reasoning → Reasoning for agents, planning and code +2
Train for Truth, Keep the Skills: Binary Retrieval-Augmented Reward Mitigates HallucinationsRetrieval-augmented generation → Hallucination detection in RAG
Can Image-To-Video Models Simulate Pedestrian Dynamics?World models → Video-generation world models and simulators +1
Rethinking Search: A Study of University Students' Perspectives on Using LLMs and Traditional Search Engines in Academic Problem SolvingLLM reasoning → Reasoning benchmarks and evaluation
MT-Video-Bench: A Holistic Video Understanding Benchmark for Evaluating Multimodal LLMs in Multi-Turn DialoguesMultimodal LLMs / vision-language models → Video understanding with MLLMs
Bi-Level Reinforcement Learning Pathway for Sim-to-Real OptimalityDeep reinforcement learning methods → RL for robot locomotion and manipulation
GAS: Improving Discretization of Diffusion ODEs via Generalized Adversarial SolverDiffusion models → Diffusion foundations and sampling
Efficient Algorithms for Mitigating Uncertainty and Risk in Reinforcement LearningDeep reinforcement learning methods → RL theory and sample complexity
CrossGuard: Safeguarding MLLMs against Joint-Modal Implicit Malicious AttacksLLM reasoning → Reasoning for agents, planning and code +2
Multilingual Text-to-Image Person Retrieval via Bidirectional Relation Reasoning and AligningLLM reasoning → Multimodal reasoning
Intelligent Communication Mixture-of-Experts Boosted-Medical Image Segmentation Foundation ModelMixture-of-experts → Medical and biomedical MoE
Qomhra: A Bilingual Irish and English Large Language ModelLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Frugal Federated Learning for Violence Detection: A Comparison of LoRA-Tuned VLMs and Personalized CNNsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
RESample: A Robust Data Augmentation Framework via Exploratory Sampling for Robotic ManipulationVision-language-action models → Failure recovery and robustness +1
SARSteer: Safeguarding Large Audio-Language Models via Safe-Ablated Refusal SteeringLLM safety and security → Safety alignment, refusal and guardrails
Forget to Know, Remember to Use: Context-Aware Unlearning for Large Language ModelsLLM safety and security → Machine unlearning
Non-asymptotic error bounds for probability flow ODEs under weak log-concavityDiffusion models → Diffusion foundations and sampling
ShapeCraft: LLM Agents for Structured, Textured and Interactive 3D ModelingLLM agents and tool use → Multimodal and embodied agents
LexChain: Modeling Legal Reasoning Chains for Chinese Tort Case AnalysisLLM reasoning → Reasoning for agents, planning and code
Reasoning Distillation and Structural Alignment for Improved Code GenerationLLM reasoning → Reasoning for agents, planning and code +1
MERIT: Modular Framework for Multimodal Misinformation Detection with Web-Grounded ReasoningLLM reasoning → Multimodal reasoning +1
Towards a Practical Understanding of Lagrangian Methods in Safe Reinforcement LearningDeep reinforcement learning methods → Safe and constrained RL
Language Confusion Gate: Language-Aware Decoding Through Model Self-DistillationKnowledge distillation → Knowledge distillation for LLMs
OncoReason: Structuring Clinical Reasoning in LLMs for Robust and Interpretable Survival PredictionLLM reasoning → Clinical and biomedical applications
Plasma Shape Control via Zero-shot Generative Reinforcement LearningDeep reinforcement learning methods → RL for energy, networks and operations
MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal Prostate MRI SegmentationState-space sequence models (S4, Mamba) → Mamba for medical image analysis
Mitigating Clever Hans Strategies in Image Classifiers through Generating CounterexamplesKnowledge distillation → General knowledge distillation methods
Cybersecurity AI: Evaluating Agentic Cybersecurity in Attack/Defense CTFsLLM agents and tool use → Agent threats, guardrails and governance
Context-Aware Pseudo-Label Scoring for Zero-Shot Video SummarizationLLM reasoning → Multimodal reasoning
Deep Self-Evolving ReasoningLLM reasoning → Test-time compute and efficient thinking +1
I-RAVEN-X: Benchmarking Generalization and Robustness of Analogical and Mathematical Reasoning in Large Language and Reasoning ModelsLLM reasoning → Mathematical reasoning
Empowering Real-World: A Survey on the Technology, Practice, and Evaluation of LLM-driven Industry AgentsLLM reasoning → Reasoning for agents, planning and code +1
ReXMoE: Reusing Experts with Minimal Overhead in Mixture-of-ExpertsMixture-of-experts → MoE large language models
SparseWorld: A Flexible, Adaptive, and Efficient 4D Occupancy World Model Powered by Sparse and Dynamic QueriesWorld models → World models for autonomous driving
Certified Self-Consistency: Statistical Guarantees and Test-Time Training for Reliable Reasoning in LLMsLLM reasoning → Test-time compute and efficient thinking
From Spatial to Actions: Grounding Vision-Language-Action Model in Spatial Foundation PriorsLLM reasoning → Spatial and embodied reasoning +1
Agentic Reinforcement Learning for Search Misaligns Instruction-TuningLLM agents and tool use → Web, search, and research agents
Navigating the Alignment-Calibration Trade-off: A Pareto-Superior Frontier via Model MergingLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Diffusion Models as Dataset Distillation PriorsDiffusion models → Diffusion foundations and sampling
Conditional Diffusion Modeling with Attention for Probabilistic Battery Capacity Prediction under Real-World ConditionDiffusion models → Flow matching, time series and science
Finite-Time Bounds for Average-Reward Fitted Q-IterationOffline RL and imitation learning from data → Offline RL algorithms
The Atomic Instruction Gap: Instruction-Tuned LLMs Struggle with Simple, Self-Contained DirectivesLLM reasoning → Reasoning benchmarks and evaluation

All 404