Papers

New on arXiv

Wed 21 Jan 2026

349 papers

XInsight: Integrative Stage-Consistent Psychological Counseling Support Agents for Digital Well-BeingLLM agents and tool use → Medical and scientific domain agents
Enhancing Tool Calling in LLMs with the International Tool Calling DatasetLLM agents and tool use → Tool learning and function calling
The Perplexity Paradox: Why Code Compresses Better Than Math in LLM PromptsLLM reasoning → Mathematical reasoning +1
Agentic AI for Commercial Insurance Underwriting with Adversarial Self-CritiqueLLM agents and tool use → Financial trading and forecasting agents
Retrieval-Augmented Self-Taught Reasoning Model with Adaptive Chain-of-Thought for ASR Named Entity CorrectionLLM reasoning → Audio-language model reasoning +1
Mind the Gap No More: Achieving Zero-Gap Multimodal Integration via One TokenizerMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
From Biased Chatbots to Biased Agents: Examining Role Assignment Effects on LLM Agent RobustnessLLM agents and tool use → Social simulation and multi-agent debate
Mechanistic Interpretability for Large Language Model Alignment: Progress, Challenges, and Future DirectionsLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
VibeTensor: System Software for Deep Learning, Fully Generated by AI AgentsLLM agents and tool use → Software engineering coding agents
QUAIL: Quantization Aware Unlearning for Mitigating Misinformation in LLMsLLM safety and security → Machine unlearning
From Generative Engines to Actionable Simulators: The Imperative of Physical Grounding in World ModelsWorld models → World models for language agents
Securing LLM-as-a-Service for Small Businesses: An Industry Case Study of a Distributed Chatbot Deployment PlatformRetrieval-augmented generation → Privacy and security of RAG +1
TransportAgents: a multi-agents LLM framework for traffic accident severity predictionLLM agents and tool use → Agent threats, guardrails and governance
Low-Dimensional Adaptation of Rectified Flow: A Diffusion and Stochastic Localization PerspectiveDiffusion models → Flow matching, time series and science
MARS: Unleashing the Power of Speculative Decoding via Margin-Aware VerificationEfficient LLM inference and serving → Speculative decoding
Tracking the Limits of Knowledge Propagation: How LLMs Fail at Multi-Step Reasoning with Conflicting KnowledgeLLM reasoning → Reasoning benchmarks and evaluation
Multi-Persona Thinking for Bias Mitigation in Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation
MiRAGE: A Multiagent Framework for Generating Multimodal Multihop Question-Answer Dataset for RAG EvaluationRetrieval-augmented generation → Agentic RAG, agents and memory
An LLM-Agnostic, MAVLink-Based Drone Command and Control Interface and Agentic Harness Using the Model Context ProtocolLLM agents and tool use → Multimodal and embodied agents
Martingale Foresight Sampling: A Principled Approach to Inference-Time LLM DecodingLLM reasoning → Test-time compute and efficient thinking
Reliability by design: quantifying and eliminating fabrication risk in LLMs. From generative to consultative AI: a comparative analysis in the legal domain and lessons for high-stakes knowledge basesRetrieval-augmented generation → Legal and financial RAG
Learning from Synthetic Data: Limitations of ERMSynthetic data and self-improvement for LLMs → LLM synthetic data generation
Chunking, Retrieval, and Re-ranking: An Empirical Evaluation of RAG Architectures for Policy Document Question AnsweringRetrieval-augmented generation → Legal and financial RAG
DevPrompt: Deviation-Based Prompt Learning for One-Normal ShotImage Anomaly DetectionMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
CASL: Concept-Aligned Sparse Latents for Interpreting Diffusion ModelsDiffusion models → Image generation, text-to-image and editing
Exploring Implicit Perspectives on Autism in Large Language Models Through Multi-Agent SimulationsLLM agents and tool use → Social simulation and multi-agent debate
Ambient Dataloops: Generative Models for Dataset RefinementDiffusion models → Diffusion foundations and sampling
CURE: Curriculum-guided Multi-task Training for Reliable Anatomy Grounded Report GenerationMultimodal LLMs / vision-language models → Medical vision-language models
Evaluating Multimodal Large Language Models for Heterogeneous Face RecognitionMultimodal LLMs / vision-language models → Social bias and fairness in VLMs
Improving MoE Compute Efficiency by Composing Weight and Data SparsityMixture-of-experts → Vision mixture-of-experts
Q-Probe: Scaling Image Quality Assessment to High Resolution via Context-Aware Agentic ProbingMultimodal LLMs / vision-language models → Image editing and quality assessment +2
Towards Understanding Best Practices for Quantization of Vision-Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency
Iterative Refinement Improves Compositional Image GenerationMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Walk through Paintings: Egocentric World Models from Internet PriorsWorld models → Embodied and robot world models +1
MolecularIQ: Characterizing Chemical Reasoning Capabilities Through Symbolic Verification on Molecular GraphsLLM reasoning → Clinical and biomedical applications
FlowSSC: Universal Generative Monocular Semantic Scene Completion via One-Step Latent DiffusionDiffusion models → 3D, point cloud and scene generation
When Agents Fail: A Comprehensive Study of Bugs in LLM Agents with Automated LabelingLLM agents and tool use → Software engineering coding agents
PROGRESSLM: Towards Progress Reasoning in Vision-Language ModelsLLM reasoning → Multimodal reasoning +1
ScenDi: 3D-to-2D Scene Diffusion Cascades for Urban GenerationDiffusion models → 3D, point cloud and scene generation
LangForce: Bayesian Decomposition of Vision Language Action Models via Latent Action QueriesVision-language-action models → Failure recovery and robustness
Where Do AI Coding Agents Fail? An Empirical Study of Failed Agentic Pull Requests in GitHubLLM agents and tool use → Software engineering coding agents
The Flexibility Trap: Rethinking the Value of Arbitrary Order in Diffusion Language ModelsLLM reasoning → Diffusion reasoning and test-time scaling +1
Retrieval-Augmented Agentic Rubric Generation for Reliable Medical Response EvaluationRetrieval-augmented generation → Medical and biomedical RAG +1
Knowledge Graphs are Implicit Reward Models: Path-Derived Signals Enable Compositional ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning +1
Outcome-Based RL Provably Leads Transformers to Reason, but Only With the Right DataLLM reasoning → Prompting and chain-of-thought reasoning
How to Build AI Agents by Augmenting LLMs with Codified Human Expert Domain Knowledge? A Software Engineering FrameworkLLM agents and tool use → Software engineering coding agents
CLEANER: Self-Purified Trajectories Boost Agentic Reinforcement LearningLLM agents and tool use → Reinforcement learning for LLM agents
RAG-GFM: Overcoming In-Memory Bottlenecks in Graph Foundation Models via Retrieval-Augmented GenerationRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
Emerging from Ground: Addressing Intent Deviation in Tool-Using Agents via Deriving Real Calls into Virtual TrajectoriesLLM agents and tool use → Tool learning and function calling
From Who They Are to How They Act: Behavioral Traits in Generative Agent-Based Models of Social MediaLLM agents and tool use → Social simulation and multi-agent debate
Auditing Language Model Unlearning via Information DecompositionLLM safety and security → Machine unlearning
An Agentic Operationalization of DISARM for FIMI Investigation on Social MediaLLM agents and tool use → Agent threats, guardrails and governance
Multi-Agent Constraint Factorization Reveals Latent Invariant Solution StructureLLM agents and tool use → General LLM agent frameworks and benchmarks
The Why Behind the Action: Unveiling Internal Drivers via Agentic AttributionLLM agents and tool use → Software engineering coding agents
SmartOracle -- An Agentic Approach to Mitigate Noise in Differential OraclesLLM agents and tool use → Software engineering coding agents
Beyond Factual Accuracy: Evaluating Global Reasoning Integrity in RAG Systems with LogicScoreLLM reasoning → Knowledge- and tool-augmented reasoning +1
Deep Leakage with Generative Flow Matching DenoiserDiffusion models → Flow matching, time series and science
Game-Theoretic Lens on LLM-based Multi-Agent SystemsLLM reasoning → Reasoning for agents, planning and code +1
HyperNet-Adaptation for Diffusion-Based Test Case GenerationDiffusion models → Diffusion foundations and sampling
Mixture-of-Experts Models in Vision: Routing, Optimization, and GeneralizationMixture-of-experts → Vision mixture-of-experts
LiViBench: An Omnimodal Benchmark for Interactive Livestream Video UnderstandingMultimodal LLMs / vision-language models → Video understanding with MLLMs
Unified Multi-Dataset Training for TBPSMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
InstructTime++: Time Series Classification with Multimodal Language Modeling via Implicit Feature EnhancementMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Towards Holistic Modeling for Video Frame Interpolation with Auto-regressive Diffusion TransformersDiffusion models → Video generation
CorpusQA: A 10 Million Token Benchmark for Corpus-Level Analysis and ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning
What Should I Cite? A RAG Benchmark for Academic Citation PredictionRetrieval-augmented generation → RAG robustness and adaptive retrieval
TIDAL: Temporally Interleaved Diffusion and Action Loop for High-Frequency VLA ControlVision-language-action models → Efficient VLA inference and adaptation
Vision-Language Models on the Edge for Real-Time Robotic PerceptionLLM reasoning → Spatial and embodied reasoning +1
CodeDelegator: Mitigating Context Pollution via Role Separation in Code-as-Action AgentsLLM agents and tool use → Software engineering coding agents
Language-Coupled Reinforcement Learning for Multilingual Retrieval-Augmented GenerationRetrieval-augmented generation → Agentic RAG, agents and memory
What Makes Low-Bit Quantization-Aware Training Work for Reasoning LLMs? A Systematic StudyLLM reasoning → Test-time compute and efficient thinking
MTFlow: Time-Conditioned Flow Matching for Microtubule Segmentation in Noisy Microscopy ImagesDiffusion models → Flow matching, time series and science
Measuring and Aligning Abstraction in Vision-Language Models with Medical TaxonomiesMultimodal LLMs / vision-language models → Medical vision-language models
UBATrack: Spatio-Temporal State Space Model for General Multi-Modal TrackingState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
Robustness of Mixtures of Experts to Feature NoiseMixture-of-experts → MoE theory and gating functions
Synthetic Data Augmentation for Multi-Task Chinese Porcelain Classification: A Stable Diffusion ApproachDiffusion models → Image generation, text-to-image and editing
CI4A: Semantic Component Interfaces for Agents Empowering Web AutomationLLM agents and tool use → Web, search, and research agents
Reconstruction-Anchored Diffusion Model for Text-to-Motion GenerationDiffusion models → Video generation
Does medical specialization of VLMs enhance discriminative power?: A comprehensive investigation through feature distribution analysisMultimodal LLMs / vision-language models → Medical vision-language models
Mechanism Shift During Post-training from Autoregressive to Masked Diffusion Language ModelsDiffusion models → Discrete and text diffusion
ReinPath: A Multimodal Reinforcement Learning Approach for PathologyLLM reasoning → Clinical and biomedical applications
Render-of-Thought: Rendering Textual Chain-of-Thought as Images for Visual Latent ReasoningLLM reasoning → Multimodal reasoning +1
Safeguarding Facial Identity against Diffusion-based Face Swapping via Cascading Pathway DisruptionDiffusion models → Robustness, privacy and adversarial aspects
DeepMoLM: Leveraging Visual and Geometric Structural Information for Molecule-Text ModelingMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video UnderstandingMultimodal LLMs / vision-language models → Video understanding with MLLMs +1
Typhoon OCR: Open Vision-Language Model For Thai Document ExtractionMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
PCL-Reasoner-V1.5: Advancing Math Reasoning with Offline Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning +1
Unified Multimodal and Multilingual Retrieval via Multi-Task Learning with NLU IntegrationMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Drive-P2D: A Progressive Perception-to-Decision Benchmark for VLMs in Autonomous DrivingEnd-to-end autonomous driving → VLMs for autonomous driving
DARL: Encouraging Diverse Answers for General Reasoning without VerifiersLLM reasoning → Reinforcement learning for reasoning
AdaTIR: Adaptive Tool-Integrated Reasoning via Difficulty-Aware Policy OptimizationLLM reasoning → Reasoning for agents, planning and code +1
CoScale-RL: Efficient Post-Training by Co-Scaling Data and ComputationLLM reasoning → Reinforcement learning for reasoning
Re-understanding Graph Unlearning through MemorizationLLM safety and security → Graph unlearning
Gaming the Judge: Unfaithful Chain-of-Thought Can Undermine Agent EvaluationLLM reasoning → Reasoning for agents, planning and code +1
IB-GRPO: Aligning LLM-based Learning Path Recommendation with Educational Objectives via Indicator-Based Group Relative Policy OptimizationLLM reasoning → LLM reasoning for recommendation
FARE: Fast-Slow Agentic Robotic ExplorationLLM agents and tool use → Reinforcement learning for LLM agents
INFA-Guard: Mitigating Malicious Propagation via Infection-Aware Safeguarding in LLM-Based Multi-Agent SystemsLLM agents and tool use → Agent threats, guardrails and governance
Query-Efficient Agentic Graph Extraction Attacks on GraphRAG SystemsLLM reasoning → Knowledge- and tool-augmented reasoning +2
NeuroFilter: Activation-Based Guardrails for Privacy-Conscious LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
Say Anything but This: When Tokenizer Betrays Reasoning in LLMsLLM reasoning → Reasoning benchmarks and evaluation

All 349