Papers

New on arXiv

Wed 14 Jan 2026

343 papers

AdaptDiff: Adaptive Guidance in Diffusion Models for Diverse and Identity-Consistent Face Synthesis (Student Abstract)Diffusion models → Image generation, text-to-image and editing
DSL-R1: From SQL to DSL for Training Retrieval Agents across Structured and Unstructured Data with Reinforcement LearningDeep reinforcement learning methods → RL training of language and agent models
A Survey of Agent Memory in the Second Half: Towards Self-Evolving and Long-Horizon AgentsLLM agents and tool use → Agent memory systems +1
CAST: Character-and-Scene Episodic Memory for AgentsLLM agents and tool use → Agent memory systems
Relevance-aware Multi-context Contrastive Decoding for Retrieval-augmented Visual Question AnsweringMultimodal LLMs / vision-language models → Document understanding and multimodal RAG +1
STEMVerse: A Dual-Axis Diagnostic Framework for STEM Reasoning in Large Language ModelsLLM reasoning → Mathematical reasoning
The Hypocrisy Gap: Quantifying Divergence Between Internal Belief and Chain-of-Thought Explanation via Sparse AutoencodersLLM reasoning → Prompting and chain-of-thought reasoning +1
SafeTalkCoach: Diversity-Driven Multi-Agent Simulation for Parent-Teen Health ConversationsLLM agents and tool use → Medical and scientific domain agents
Rethinking LLM-Driven Heuristic Design: Generating Efficient and Specialized Solvers via Dynamics-Aware OptimizationLLM reasoning → Test-time compute and efficient thinking
MathMixup: Boosting LLM Mathematical Reasoning with Difficulty-Controllable Data Synthesis and Curriculum LearningLLM reasoning → Mathematical reasoning +1
RAM-SD: Retrieval-Augmented Multi-agent framework for Sarcasm DetectionRetrieval-augmented generation → Agentic RAG, agents and memory +1
Aeon: High-Performance Neuro-Symbolic Memory Management for Long-Horizon LLM AgentsLLM reasoning → Reasoning for agents, planning and code +1
Project Aletheia: Verifier-Guided Distillation of Backtracking for Small Language ModelsLLM reasoning → Test-time compute and efficient thinking
Chain-of-Memory: Lightweight Memory Construction with Dynamic Evolution for LLM AgentsLLM agents and tool use → Agent memory systems
Predicting When to Trust Vision-Language Models for Spatial ReasoningLLM reasoning → Spatial and embodied reasoning +1
Mixture of Distributions Matters: Dynamic Sparse Attention for Efficient Video Diffusion TransformersDiffusion models → Video generation
Evaluating Self-Correcting Vision Agents Through Quantitative and Qualitative MetricsLLM agents and tool use → Multimodal and embodied agents
When Rules Fall Short: Agent-Driven Discovery of Emerging Content Issues in Short Video PlatformsLLM agents and tool use → Multimodal and embodied agents
Beyond Accuracy: Evaluating Grounded Visual Evidence in Thinking with ImagesLLM reasoning → Multimodal reasoning +1
KG-ViP: Bridging Knowledge Grounding and Visual Perception in Multi-modal LLMs for Visual Question AnsweringMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Compress to Focus: Efficient Coordinate Compression for Policy Optimization in Multi-Turn GUI AgentsLLM agents and tool use → GUI and mobile agents +1
A one-step generation model with a Single-Layer Transformer: Layer number re-distillation of FreeFlowDiffusion models → Flow matching, time series and science
Semantic Differentiation for Tackling Challenges in Watermarking Low-Entropy Constrained Generation OutputsLLM safety and security → Watermarking generated content and models
Agents at Risk: How Users Unwittingly Undermine LLM SafetyLLM agents and tool use → Agent threats, guardrails and governance +1
From SERPs to Agents: A Platform for Comparative Studies of Information InteractionLLM agents and tool use → GUI and mobile agents
Hallucination Detection and Mitigation in Large Language ModelsLLM reasoning → Hallucination detection and mitigation
In-Browser Agents for Search AssistanceLLM agents and tool use → Web, search, and research agents
CaMeLs Can Use Computers Too: System-level Security for Computer Use AgentsLLM agents and tool use → Agent threats, guardrails and governance +1
Continuum Memory Architectures for Long-Horizon LLM AgentsRetrieval-augmented generation → Agentic RAG, agents and memory +1
Beyond Rule-Based Workflows: An Information-Flow-Orchestrated Multi-Agents Paradigm via Agent-to-Agent Communication from CORALLLM agents and tool use → Software engineering coding agents
Transition Matching Distillation for Fast Video GenerationDiffusion models → Video generation
MedVL-SAM2: A unified 3D medical vision-language model for multimodal reasoning and prompt-driven segmentationLLM reasoning → Clinical and biomedical applications +1
A Scoping Review of the Ethical Perspectives on Anthropomorphising Large Language Model-Based Conversational AgentsLLM agents and tool use → Social simulation and multi-agent debate
OUTLINEFORGE: Hierarchical Reinforcement Learning with Explicit States for Scientific WritingDeep reinforcement learning methods → RL training of language and agent models
Thinking Long, but Short: Stable Sequential Test-Time Scaling for Large Reasoning ModelsLLM reasoning → Test-time compute and efficient thinking +1
NanoSD: Edge Efficient Foundation Model for Real Time Image RestorationDiffusion models → Inverse problems and image restoration
LLM-Based Agentic Systems for Software Engineering: Challenges and OpportunitiesLLM agents and tool use → Software engineering coding agents
Improving Chain-of-Thought for Logical Reasoning via Attention-Aware InterventionLLM reasoning → Test-time compute and efficient thinking
TimeSAE: Causal Sparse Decoding for Faithful Explanations of Black-Box Time Series ModelsMechanistic interpretability → Interpreting time-series and weather models
PCN-Rec: Agentic Proof-Carrying Negotiation for Reliable Governance-Constrained RecommendationLLM agents and tool use → Agentic recommender systems
GUI-Eyes: Tool-Augmented Perception for Visual Grounding in GUI AgentsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +2
Investigating Tool-Memory Conflicts in Tool-Augmented LLMsLLM agents and tool use → Tool learning and function calling
Fast-ThinkAct: Efficient Vision-Language-Action Reasoning via Verbalizable Latent PlanningVision-language-action models → Reasoning and chain-of-thought in VLA
Value-Aware Numerical Representations for Transformer Language ModelsLLM reasoning → Mathematical reasoning
ShortCoder: Knowledge-Augmented Syntax Optimization for Token-Efficient Code GenerationEfficient LLM inference and serving → Speculative decoding
Efficient Camera-Controlled Video Generation of Static Scenes via Sparse Diffusion and 3D RenderingDiffusion models → Video generation
LLMs can Compress LLMs: Adaptive Pruning by AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
DeepResearchEval: An Automated Framework for Deep Research Task Construction and Agentic EvaluationLLM agents and tool use → Web, search, and research agents
Automating Supply Chain Disruption Monitoring via an Agentic AI ApproachLLM agents and tool use → Social simulation and multi-agent debate
Collaborative Multi-Agent Test-Time Reinforcement Learning for ReasoningLLM reasoning → Reasoning for agents, planning and code +1
LiteEmbed: Adapting CLIP to Rare ClassesMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
PersonalAlign: Hierarchical Implicit Intent Alignment for Personalized GUI Agent with Long-Term User-Centric RecordsLLM agents and tool use → GUI and mobile agents
Large-Scale Optimization Model Auto-Formulation: Harnessing LLM Flexibility via Structured WorkflowLLM agents and tool use → General LLM agent frameworks and benchmarks
The Promptware Kill Chain: How Prompt Injections Gradually Evolved Into a Multistep Malware Delivery MechanismLLM safety and security → Prompt injection and agent security
A Mechanistic Perspective and Circuit-Guided Difficulty Metric for UnlearningLLM safety and security → Machine unlearning
DPWriter: Reinforcement Learning with Diverse Planning Branching for Creative WritingDeep reinforcement learning methods → RL training of language and agent models
Energy-Entropy Regularization: The True Power of Minimal Looped TransformersLLM reasoning → Prompting and chain-of-thought reasoning
OpenVoxel: Training-Free Grouping and Captioning Voxels for Open-Vocabulary 3D Scene UnderstandingMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Trustworthy Longitudinal Brain MRI Completion: A Deformation-Based Approach with KAN-Enhanced Diffusion ModelDiffusion models → Medical imaging and segmentation
Omni-R1: Towards the Unified Generative Paradigm for Multimodal ReasoningLLM reasoning → Multimodal reasoning +1
TEMPO: A Realistic Multi-Domain Benchmark for Temporal Reasoning-Intensive RetrievalLLM reasoning → Knowledge- and tool-augmented reasoning
CLARE: Continual Learning for Vision-Language-Action Models via Autonomous Adapter Routing and ExpansionVision-language-action models → Reinforcement learning and post-training
What Do LLM Agents Know About Their World? Task2Quiz: A Paradigm for Studying Environment UnderstandingLLM agents and tool use → General LLM agent frameworks and benchmarks
FairGU: Fairness-aware Graph Unlearning in Social NetworksLLM safety and security → Graph unlearning
EvoFSM: Controllable Self-Evolution for Deep Research with Finite State MachinesLLM agents and tool use → General LLM agent frameworks and benchmarks
Dissecting Judicial Reasoning in U.S. Copyright Damage AwardsLLM reasoning → Reasoning for agents, planning and code
MAD: Motion Appearance Decoupling for efficient Driving World ModelsWorld models → World models for autonomous driving +2
Improving Symbolic Translation of Language Models for Logical ReasoningLLM reasoning → Reasoning benchmarks and evaluation
DepRadar: Agentic Coordination for Context Aware Defect Impact Analysis in Deep Learning LibrariesLLM agents and tool use → Software engineering coding agents
SC-MAS: Constructing Cost-Efficient Multi-Agent Systems with Edge-Level Heterogeneous CollaborationLLM agents and tool use → General LLM agent frameworks and benchmarks
Video-MSR: Benchmarking Multi-hop Spatial Reasoning Capabilities of MLLMsLLM reasoning → Spatial and embodied reasoning +1
Speech-Hands: A Self-Reflection Voice Agentic Approach to Speech Recognition and Audio Reasoning with Omni PerceptionLLM agents and tool use → General LLM agent frameworks and benchmarks
SEEK: Steering LLM Reasoning for RAG via Internal Reasoning SketchesLLM reasoning → Knowledge- and tool-augmented reasoning +1
AI-NativeBench: An Open-Source White-Box Agentic Benchmark Suite for AI-Native SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
SLAM-LLM: A Modular, Open-Source Multimodal Large Language Model Framework and Best Practice for Speech, Language, Audio and Music ProcessingMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Long-term Task-oriented Agent: Proactive Long-term Intent Maintenance in Dynamic EnvironmentsLLM agents and tool use → General LLM agent frameworks and benchmarks
ReflexDiffusion: Reflection-Enhanced Trajectory Planning for High-lateral-acceleration Scenarios in Autonomous DrivingEnd-to-end autonomous driving → Imitation-learning end-to-end planners
GeoRA: Geometry-Aware Low-Rank Adaptation for RLVRLLM reasoning → Reinforcement learning for reasoning
Monte-Carlo Tree Search with Neural Network Guidance for Lane-Free Autonomous DrivingEnd-to-end autonomous driving → Imitation-learning end-to-end planners
See More, Store Less: Memory-Efficient Resolution for Video Moment RetrievalMultimodal LLMs / vision-language models → Video understanding with MLLMs
Research on Piano Timbre Transformation System Based on Diffusion ModelDiffusion models → Audio, speech and music
SpatialJB: How Text Distribution Art Becomes the "Jailbreak Key" for LLM GuardrailsLLM safety and security → Jailbreaking
Multi-Modal LLM based Image Captioning in ICT: Bridging the Gap Between General and Industry DomainMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
MACRO-LLM: LLM-Empowered Multi-Agent Collaborative Reasoning under Spatiotemporal Partial ObservabilityLLM reasoning → Spatial and embodied reasoning +1
Policy-Based Reinforcement Learning with Action Masking for Dynamic Job Shop Scheduling under Uncertainty: Handling Random Arrivals and Machine FailuresDeep reinforcement learning methods → RL for energy, networks and operations
Blue Teaming Function-Calling AgentsLLM agents and tool use → Agent threats, guardrails and governance
Enhancing Spatial Reasoning in Large Language Models for Metal-Organic Frameworks Structure PredictionLLM reasoning → Clinical and biomedical applications
STaR: Sensitive Trajectory Regulation for Unlearning in Large Reasoning ModelsLLM reasoning → Reasoning for agents, planning and code +1
ReGraM: Region-First Knowledge Graph Reasoning for Medical Question AnsweringLLM reasoning → Clinical and biomedical applications
A3-Bench: Benchmarking Memory-Driven Scientific Reasoning via Anchor and Attractor ActivationLLM reasoning → Test-time compute and efficient thinking
RISER: Orchestrating Latent Reasoning Skills for Adaptive Activation SteeringLLM reasoning → Test-time compute and efficient thinking +1
Efficient Paths and Dense Rewards: Probabilistic Flow Reasoning for Large Language ModelsLLM reasoning → Test-time compute and efficient thinking
MAXS: Meta-Adaptive Exploration with LLM AgentsLLM reasoning → Reasoning for agents, planning and code +1
PhyRPR: Training-Free Physics-Constrained Video GenerationDiffusion models → Video generation
RIFT: Repurposing Negative Samples via Reward-Informed Fine-TuningLLM alignment / RLHF / preference optimisation → LLM preference optimization methods +1
When to Trust: A Causality-Aware Calibration Framework for Accurate Knowledge Graph Retrieval-Augmented GenerationRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
DSA-Tokenizer: Disentangled Semantic-Acoustic Tokenization via Flow Matching-based Hierarchical FusionDiffusion models → Audio, speech and music
Knowledge-Embedded and Hypernetwork-Guided Few-Shot Substation Meter Defect Image Generation MethodDiffusion models → Flow matching, time series and science
Reward Learning through Ranking Mean Squared ErrorLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
GIFT: Reconciling Post-Training Objectives via Variational Finite-Temperature Gibbs InitializationLLM reasoning → Reinforcement learning for reasoning +1

All 343