Papers

New on arXiv

Mon 23 Mar 2026

435 papers

SOLAR: A Self-Optimizing Open-Ended Autonomous Agent for Lifelong Learning and Continual AdaptationLLM agents and tool use → General LLM agent frameworks and benchmarks
How to Fine-Tune a Reasoning Model? A Teacher-Student Cooperation Framework to Synthesize Student-Consistent SFT DataLLM reasoning → Prompting and chain-of-thought reasoning +1
NuHF Claw: A Risk Constrained Cognitive Agent Framework for Human Centered Procedure Support in Digital Nuclear Control RoomsLLM agents and tool use → Agent threats, guardrails and governance
Alignment as Institutional Design: From Behavioral Correction to Transaction Structure in Intelligent SystemsLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
IPSL-AID: Generative Diffusion Models for Climate Downscaling from Global to Regional ScalesDiffusion models → Flow matching, time series and science
Evolutionary Warm-Starts for Reinforcement Learning in Industrial Continuous ControlDeep reinforcement learning methods → RL for energy, networks and operations
From Diffusion to Flow: Efficient Motion Generation in MotionGPT3Diffusion models → Video generation
Steering Sparse Autoencoder Latents to Control Dynamic Head Pruning in Vision Transformers (Student Abstract)Mechanistic interpretability → Interpreting vision and multimodal models
Do Multilingual VLMs Reason Equally? A Cross-Lingual Visual Reasoning Audit for Indian LanguagesLLM reasoning → Reasoning benchmarks and evaluation +1
Language-Conditioned World Modeling for Visual NavigationWorld models → World models for navigation
CAPTCHA Solving for Native GUI Agents: Automated Reasoning-Action Data Generation and Self-Corrective TrainingLLM reasoning → Multimodal reasoning +2
Mixture of Demonstrations for Textual Graph Understanding and Question AnsweringRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG +1
AwesomeLit: Towards Hypothesis Generation with Agent-Supported Literature ResearchLLM agents and tool use → Multimodal and embodied agents
Q-Tacit: Image Quality Assessment via Latent Visual ReasoningLLM reasoning → Multimodal reasoning +1
Graph-Aware Late Chunking for Retrieval-Augmented Generation in Biomedical LiteratureRetrieval-augmented generation → Medical and biomedical RAG
To Agree or To Be Right? The Grounding-Sycophancy Tradeoff in Medical Vision-Language ModelsMultimodal LLMs / vision-language models → Medical vision-language models
Bridging the Know-Act Gap via Task-Level Autoregressive ReasoningLLM reasoning → Test-time compute and efficient thinking +1
Language Models Can Explain Visual Features via SteeringMechanistic interpretability → Interpreting vision and multimodal models
A vision-language model and platform for temporally mapping surgery from videoMultimodal LLMs / vision-language models → Medical vision-language models
Lie to Me: How Faithful Is Chain-of-Thought Reasoning in Reasoning Models?LLM reasoning → Prompting and chain-of-thought reasoning
STRIATUM-CTF: A Protocol-Driven Agentic Framework for General-Purpose CTF SolvingLLM agents and tool use → Agent threats, guardrails and governance
TrustTrade: Human-Inspired Selective Consensus Reduces Decision Uncertainty in LLM Trading AgentsLLM agents and tool use → Financial trading and forecasting agents
Privacy-Preserving Reinforcement Learning from Human Feedback via Decoupled Reward ModelingLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment +1
Ego2Web: A Web Agent Benchmark Grounded in Egocentric VideosLLM agents and tool use → Multimodal and embodied agents
GraphRAG for Engineering Diagrams: ChatP&ID Enables LLM Interaction with P&IDsRetrieval-augmented generation → Knowledge-graph and structured-knowledge RAG
Learning Sidewalk Autopilot from Multi-Scale Imitation with Corrective Behavior ExpansionOffline RL and imitation learning from data → Imitation and offline RL for driving
Sketch2CT: Multimodal Diffusion for Structure-Aware 3D Medical Volume GenerationDiffusion models → Medical imaging and segmentation
Tiny Inference-Time Scaling with Latent VerifiersLLM reasoning → Diffusion reasoning and test-time scaling
Model Context Protocol Threat Modeling and Analyzing Vulnerabilities to Prompt Injection with Tool PoisoningLLM safety and security → Prompt injection and agent security
MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion DecodingLLM reasoning → Diffusion reasoning and test-time scaling
SkillRouter: Skill Routing for LLM Agents at ScaleLLM agents and tool use → Agent skill learning and evolution
Latent Reuse in Agent Skills: Multi-modal Clone Detection at Ecosystem ScaleLLM agents and tool use → Agent skill learning and evolution
Sparse but Critical: A Token-Level Analysis of Distributional Shifts in RLVR Fine-Tuning of LLMsLLM reasoning → Reinforcement learning for reasoning
CaP-X: A Framework for Benchmarking and Improving Coding Agents for Robot ManipulationVision-language-action models → Failure recovery and robustness +1
Inference Time Policy Optimization for Offline RL with Differentiable World ModelsWorld models → Latent dynamics models and model-based RL +1
From Static Templates to Dynamic Runtime Graphs: A Survey of Workflow Optimization for LLM AgentsLLM agents and tool use → Software engineering coding agents
Learning When to Act: Interval-Aware Reinforcement Learning with Predictive Temporal StructureDeep reinforcement learning methods → Deep RL algorithms and representations
Closing the Auto-Research Loop: An AI Co-Scientist for Production Search RankingLLM agents and tool use → Web, search, and research agents
Three Creates All: You Only Sample 3 StepsDiffusion models → Diffusion foundations and sampling
When Visuals Aren't the Problem: Evaluating Vision-Language Models on Misleading Data VisualizationsMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
Reasoner-Executor-Synthesizer: Scalable Agentic Architecture with Static O(1) Context WindowLLM reasoning → Knowledge- and tool-augmented reasoning +2
MCLR: Improving Conditional Modeling via Inter-Class Likelihood-Ratio Maximization and Unifying Classifier-Free Guidance with Alignment ObjectivesDiffusion models → Diffusion foundations and sampling
Early Discoveries of Algorithmist I: Promise of Provable Algorithm Synthesis at ScaleLLM reasoning → Mathematical reasoning
WorldCache: Content-Aware Caching for Accelerated Video World ModelsWorld models → Video-generation world models and simulators +1
VideoDetective: Clue Hunting via both Extrinsic Query and Intrinsic Relevance for Long Video UnderstandingMultimodal LLMs / vision-language models → Video understanding with MLLMs
End-to-End Training for Unified Tokenization and Latent DenoisingDiffusion models → Diffusion foundations and sampling
ThinkJEPA: Empowering Latent World Models with Large Vision-Language Reasoning ModelWorld models → Embodied and robot world models +1
DualCoT-VLA: Visual-Linguistic Chain of Thought via Parallel Reasoning for Vision-Language-Action ModelsLLM reasoning → Spatial and embodied reasoning +1
3D-Layout-R1: Structured Reasoning for Language-Instructed Spatial EditingLLM reasoning → Spatial and embodied reasoning +1
The Dual Mechanisms of Spatial Variable Binding in Vision-Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Repurposing Geometric Foundation Models for Multi-view DiffusionDiffusion models → 3D, point cloud and scene generation
Decoupling Exploration and Policy Optimization: Uncertainty Guided Tree Search for Hard ExplorationDeep reinforcement learning methods → Deep RL algorithms and representations
DUO-VSR: Dual-Stream Distillation for One-Step Video Super-ResolutionDiffusion models → Inverse problems and image restoration
Confidence-Based Decoding is Provably Efficient for Diffusion Language ModelsDiffusion models → Discrete and text diffusion
MemDLM: Memory-Enhanced DLM TrainingDiffusion models → Discrete and text diffusion +1
SpatialReward: Verifiable Spatial Reward Modeling for Fine-Grained Spatial Consistency in Text-to-Image GenerationLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
Gumbel Distillation for Parallel Text GenerationEfficient LLM inference and serving → Diffusion LLM and MoE inference
Omni-WorldBench: Towards a Comprehensive Interaction-Centric Evaluation for World ModelsWorld models → Video-generation world models and simulators
Chimera: Latency- and Performance-Aware Multi-agent Serving for Heterogeneous LLMsLLM agents and tool use → Software engineering coding agents
Seeing is Improving: Visual Feedback for Iterative Text Layout RefinementMultimodal LLMs / vision-language models → Document understanding and multimodal RAG +1
Enhancing Document-Level Machine Translation via Filtered Synthetic Corpora and Two-Stage LLM AdaptationSynthetic data and self-improvement for LLMs → LLM synthetic data generation
Revisiting Quantum Code Generation: Where Should Domain Knowledge Live?Retrieval-augmented generation → RAG for code
Cross-Modal Reinforcement Learning for Navigation with Degraded Depth MeasurementsDeep reinforcement learning methods → RL for robot navigation and UAV control
MARCUS: An agentic, multimodal vision-language model for cardiac diagnosis and managementMultimodal LLMs / vision-language models → Medical vision-language models +1
VersualRL: Closed-Loop Verbal Reinforcement Learning with Visual Execution Feedback for Task-Level Robot PlanningDeep reinforcement learning methods → RL for robot locomotion and manipulation
ACPO: Counteracting Likelihood Displacement in Vision-Language Alignment with Asymmetric ConstraintsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
OpenEarth-Agent: From Tool Calling to Tool Creation for Open-Environment Earth ObservationLLM agents and tool use → Multimodal and embodied agents
DA-VAE: Plug-in Latent Compression for Diffusion via Detail AlignmentDiffusion models → Diffusion foundations and sampling
On the Direction of RLVR Updates for LLM Reasoning: Identification and ExploitationLLM reasoning → Reinforcement learning for reasoning
Automated Lemma Discovery in Agentic Program VerificationLLM agents and tool use → Educational agents and math problem solving
GSEM: Graph-based Self-Evolving Memory for Experience Augmented Clinical ReasoningLLM reasoning → Clinical and biomedical applications +1
Principled Steering via Null-space Projection for Jailbreak Defense in Vision-Language ModelsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +2
A Context Engineering Framework for Improving Enterprise AI Agents based on Digital-Twin MDPLLM agents and tool use → Reinforcement learning for LLM agents
Do World Action Models Generalize Better than VLAs? A Robustness StudyWorld models → Embodied and robot world models +1
Autoregressive vs. Masked Diffusion Language Models: A Controlled ComparisonDiffusion models → Discrete and text diffusion
Adapting Point Cloud Analysis via Multimodal Bayesian Distribution LearningMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
On the Failure of Topic-Matched Contrast Baselines in Multi-Directional Refusal AbliterationLLM safety and security → Safety alignment, refusal and guardrails
SpatialBoost: Enhancing Visual Representation through Language-Guided ReasoningLLM reasoning → Spatial and embodied reasoning
Dual-Space Knowledge Distillation with Key-Query Matching for Large Language Models with Vocabulary MismatchKnowledge distillation → Knowledge distillation for LLMs
Uncertainty-guided Compositional Alignment with Part-to-Whole Semantic Representativeness in Hyperbolic Vision-Language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
DTVI: Dual-Stage Textual and Visual Intervention for Safe Text-to-Image GenerationDiffusion models → Robustness, privacy and adversarial aspects +1
Tuning Real-World Image Restoration at Inference: A Test-Time Scaling Paradigm for Flow Matching ModelsDiffusion models → Inverse problems and image restoration
ROM: Real-time Overthinking Mitigation via Streaming Detection and InterventionLLM reasoning → Test-time compute and efficient thinking
VP-VLA: Visual Prompting as an Interface for Vision-Language-Action ModelsVision-language-action models → Reasoning and chain-of-thought in VLA
SegMaFormer: A Hybrid State-Space and Transformer Model for Efficient SegmentationState-space sequence models (S4, Mamba) → Mamba for medical image analysis
TREX: Trajectory Explanations for Multi-Objective Reinforcement LearningDeep reinforcement learning methods → RL for games and combinatorial optimization
SecureBreak -- A dataset towards safe and secure modelsLLM safety and security → Safety alignment, refusal and guardrails
Demystifying Reinforcement Learning for Long-Horizon Tool-Using Agents: A Comprehensive RecipeLLM agents and tool use → Reinforcement learning for LLM agents +1
Unified Spatiotemporal Token Compression for Video-LLMs at Ultra-Low RetentionEfficient LLM inference and serving → Visual token pruning in multimodal models
FeatDistill: A Feature Distillation Enhanced Multi-Expert Ensemble Framework for Robust AI-generated Image DetectionKnowledge distillation → Detection and dense prediction distillation
Guideline-grounded retrieval-augmented generation for ophthalmic clinical decision supportRetrieval-augmented generation → Medical and biomedical RAG
Deep Reinforcement Learning and The Tale of Two Temporal Difference ErrorsDeep reinforcement learning methods → RL theory and sample complexity
ADaFuSE: Adaptive Diffusion-generated Image and Text Fusion for Interactive Text-to-Image RetrievalDiffusion models → Image generation, text-to-image and editing
Not All Layers Are Created Equal: Adaptive LoRA Ranks for Personalized Image GenerationDiffusion models → Image generation, text-to-image and editing
P^2O: Joint Policy and Prompt OptimizationLLM reasoning → Reinforcement learning for reasoning
Manifold-Aware Exploration for Reinforcement Learning in Video GenerationLLM reasoning → Reinforcement learning for reasoning +1
Adaptive Video Distillation: Mitigating Oversaturation and Temporal Collapse in Few-Step GenerationDiffusion models → Video generation
Holistic Scaling Laws for Optimal Mixture-of-Experts Architecture OptimizationMixture-of-experts → Expert routing strategies
Climate Prompting: Generating the Madden-Julian Oscillation using Video Diffusion and Low-Dimensional ConditioningDiffusion models → Flow matching, time series and science
Reasoning or Rhetoric? An Empirical Analysis of Moral Reasoning Explanations in Large Language ModelsLLM reasoning → Reasoning benchmarks and evaluation

All 435