Papers

New on arXiv

Wed 26 Nov 2025

378 papers

MegaRAG: Multimodal Knowledge Graph-Based Retrieval Augmented GenerationRetrieval-augmented generation → Multimodal RAG
Graph-O1 : Monte Carlo Tree Search with Reinforcement Learning for Text-Attributed Graph ReasoningLLM reasoning → Knowledge- and tool-augmented reasoning +1
Towards Reasoning-Preserving Unlearning in Multimodal Large Language ModelsLLM reasoning → Reasoning for agents, planning and code +2
Efficient Multi-Adapter LLM Serving via Cross-Model KV-Cache Reuse with Activated LoRAEfficient LLM inference and serving → Serving systems and request scheduling
Reinforcement Learning for Latent-Space Thinking in LLMsLLM reasoning → Prompting and chain-of-thought reasoning
Quantum Circuit Reasoning Models: A Variational Framework for Differentiable Logical InferenceLLM reasoning → Reinforcement learning for reasoning
FAIM: Frequency-Aware Interactive Mamba for Time Series ClassificationState-space sequence models (S4, Mamba) → Time series forecasting with SSMs
GPU Memory Prediction for Multimodal Model TrainingLLM agents and tool use → Software engineering coding agents
SABER: Small Actions, Big Errors -- Safeguarding Mutating Steps in LLM AgentsLLM agents and tool use → Software engineering coding agents
AI Urban Scientist: Multi-Agent Collaborative Automation for Urban ResearchLLM agents and tool use → Multimodal and embodied agents
Optimizing Life Sciences Agents in Real-Time using Reinforcement LearningDeep reinforcement learning methods → RL training of language and agent models
Beyond Confidence: Adaptive and Coherent Decoding for Diffusion Language ModelsDiffusion models → Discrete and text diffusion
HMARK: Radioactive Multi-Bit Semantic-Latent Watermarking for Diffusion ModelsDiffusion models → Robustness, privacy and adversarial aspects +1
Exploring Automated Recognition of Instructional Activity and Discourse from Multimodal Classroom DataMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Hyper-GoalNet: Goal-Conditioned Manipulation Policy Learning with HyperNetworksDeep reinforcement learning methods → RL for robot locomotion and manipulation
A Fast and Efficient Modern BERT based Text-Conditioned Diffusion Model for Medical Image SegmentationDiffusion models → Medical imaging and segmentation
Exploring Diagnostic Prompting Approach for Multimodal LLM-based Visual Complexity Assessment: A Case Study of Amazon Search Result PagesMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
PAT3D: Physics-Augmented Text-to-3D Scene GenerationMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Prompted Policy Search: Reinforcement Learning through Linguistic and Numerical Reasoning in LLMsLLM reasoning → Reinforcement learning for reasoning
OpenTwinMap: An Open-Source Digital Twin Generator for Urban Autonomous DrivingWorld models → World models for autonomous driving +1
PathReasoning: A multimodal reasoning agent for query-based ROI navigation on whole-slide imagesLLM reasoning → Clinical and biomedical applications
From Discrete Plans to Real-World Execution: A World-Model-Driven Framework for Execution-Aware Multi-Agent Path FindingWorld models → World models for language agents
Saddle-Free Guidance: Improved On-Manifold Sampling without Labels or Additional TrainingDiffusion models → Diffusion foundations and sampling
OOCO: Latency-disaggregated Architecture for Online-Offline Co-locate LLM ServingEfficient LLM inference and serving → Serving systems and request scheduling
Towards a Foundation Model for Partial Differential Equations Across Physics DomainsLLM reasoning → Spatial and embodied reasoning
Massively Parallel Imitation Learning of Mouse Forelimb Musculoskeletal Reaching DynamicsOffline RL and imitation learning from data → Real-world robot offline RL
3MDiT: Unified Tri-Modal Diffusion Transformer for Text-Driven Synchronized Audio-Video GenerationDiffusion models → Audio, speech and music
Canvas-to-Image: Compositional Image Generation with Multimodal ControlsDiffusion models → Image generation, text-to-image and editing
TraceGen: World Modeling in 3D Trace Space Enables Learning from Cross-Embodiment VideosWorld models → Embodied and robot world models
G2VLM: Geometry Grounded Vision Language Model with Unified 3D Reconstruction and Spatial ReasoningLLM reasoning → Spatial and embodied reasoning +1
Matrix: Peer-to-Peer Multi-Agent Synthetic Data Generation FrameworkLLM agents and tool use → Software engineering coding agents +1
Agentic Learner with Grow-and-Refine Multimodal Semantic MemoryLLM reasoning → Multimodal reasoning +1
DSD: A Distributed Speculative Decoding Solution for Edge-Cloud Agile Large Model ServingEfficient LLM inference and serving → Early exit and edge inference
Escaping the Verifier: Learning to Reason via DemonstrationsLLM reasoning → Reinforcement learning for reasoning
Attention-Guided Patch-Wise Sparse Adversarial Attacks on Vision-Language-Action ModelsVision-language-action models → Safety and adversarial attacks
Multi-Crit: Benchmarking Multimodal Judges on Pluralistic Criteria-FollowingMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
EvilGenie: A Reward Hacking BenchmarkLLM alignment / RLHF / preference optimisation → Reward hacking and misspecification
Iterative GRPO: Batch-Online Multi-Turn RL via Single-Turn RLHFLLM reasoning → Reinforcement learning for reasoning +2
Qwen3-VL Technical ReportMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
Robust Spectral Watermark for Synthetic Tabular DataLLM safety and security → Watermarking generated content and models
Closed Form HJB Solution for Continuous-Time Optimal Control of a Non-Linear Input-Affine SystemDeep reinforcement learning methods → RL theory and sample complexity
MoGAN: Improving Motion Quality in Video Diffusion via Few-Step Motion Adversarial Post-TrainingDiffusion models → Video generation
Model-Based Policy Adaptation for Closed-Loop End-to-End Autonomous DrivingEnd-to-end autonomous driving → End-to-end system architecture
Learning When to Stop: Adaptive Latent Reasoning via Reinforcement LearningLLM reasoning → Test-time compute and efficient thinking
BAMAS: Structuring Budget-Aware Multi-Agent SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
RoParQ: Paraphrase-Aware Alignment of Large Language Models Towards Robustness to Paraphrased QuestionsLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
VacuumVLA: Boosting VLA Capabilities via a Unified Suction and Gripping Tool for Complex Robotic ManipulationVision-language-action models → Tactile, force-aware and humanoid VLA
MMA: A Momentum Mamba Architecture for Human Activity Recognition with Inertial SensorsState-space sequence models (S4, Mamba) → Video, motion and generative Mamba models
Seeing Twice: How Side-by-Side T2I Comparison Changes Auditing StrategiesLLM safety and security → Vision-language and text-to-image safety
E0: Enhancing Generalization and Fine-Grained Control in VLA Models via Tweedie Discrete DiffusionVision-language-action models → Efficient VLA inference and adaptation +1
Video Generation Models Are Good Latent Reward ModelsLLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment
Context-Specific Causal Graph Discovery with Unobserved Contexts: Non-Stationarity, Regimes and Spatio-Temporal PatternsLLM reasoning → Causal reasoning in LLMs
Predictive Safety Shield for Dyna-Q Reinforcement LearningDeep reinforcement learning methods → Safe and constrained RL
Pessimistic Verification for Open Ended Math QuestionsDeep reinforcement learning methods → RL training of language and agent models
Mechanistic Interpretability for Transformer-based Time Series ClassificationMechanistic interpretability → Interpreting time-series and weather models
Tool-RoCo: An Agent-as-Tool Self-organization Large Language Model Benchmark in Multi-robot CooperationLLM agents and tool use → General LLM agent frameworks and benchmarks
CanKD: Cross-Attention-based Non-local operation for Feature-based Knowledge DistillationKnowledge distillation → Detection and dense prediction distillation
Frequency-Aware Token Reduction for Efficient Vision TransformerEfficient LLM inference and serving → Token pruning and merging for ViTs
MobileI2V: Fast and High-Resolution Image-to-Video on Mobile DevicesDiffusion models → Video generation
SpatialBench: Benchmarking Multimodal Large Language Models for Spatial CognitionMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
MADRA: Multi-Agent Debate for Risk-Aware Embodied PlanningLLM agents and tool use → Agent threats, guardrails and governance
EWE: An Agentic Framework for Extreme Weather AnalysisLLM agents and tool use → Financial trading and forecasting agents
Conversational No-code, Multi-agentic Disease Module Identification and Drug Repurposing Prediction with ChatDRexLLM agents and tool use → Medical and scientific domain agents
MemFine: Memory-Aware Fine-Grained Scheduling for MoE TrainingMixture-of-experts → Efficient MoE training and inference systems
From Observation to Action: Latent Action-based Primitive Segmentation for VLA Pre-training in Industrial SettingsVision-language-action models → Human-video and latent-action pretraining
DiverseVAR: Balancing Diversity and Quality of Next-Scale Visual Autoregressive ModelsDiffusion models → Image generation, text-to-image and editing
Knowledge Distillation for Continual Learning of Biomedical Neural FieldsKnowledge distillation → Distillation for continual learning
Subjective Depth and Timescale Transformers: Learning Where and When to ComputeEfficient LLM inference and serving → Early exit and edge inference
Decentralized Shepherding of Non-Cohesive Swarms Through Cluttered Environments via Deep Reinforcement LearningDeep reinforcement learning methods → RL for robot navigation and UAV control
Text-to-SQL as Dual-State Reasoning: Integrating Adaptive Context and Progressive GenerationLLM reasoning → Knowledge- and tool-augmented reasoning
Steering Awareness: Detecting Activation Steering from WithinMechanistic interpretability → Activation steering methods
Prune4Web: DOM Tree Pruning Programming for Web AgentLLM agents and tool use → Web, search, and research agents
Understanding the Effects of Distractors on Reasoning Vision-Language ModelsLLM reasoning → Multimodal reasoning +1
Monet: Reasoning in Latent Visual Space Beyond Images and LanguageLLM reasoning → Multimodal reasoning
Multi-Agent Systems for Dataset Adaptation in Software Engineering: Capabilities, Limitations, and Future DirectionsLLM agents and tool use → Software engineering coding agents
Thinking With Bounding Boxes: Enhancing Spatio-Temporal Video Grounding via Reinforcement Fine-TuningLLM reasoning → Multimodal reasoning
Hybrid-AIRL: Enhancing Inverse Reinforcement Learning with Supervised Expert GuidanceOffline RL and imitation learning from data → Inverse reinforcement learning +1
Generating Separated Singing Vocals Using a Diffusion Model Conditioned on Music MixturesDiffusion models → Audio, speech and music
SurgMLLMBench: A Multimodal Large Language Model Benchmark Dataset for Surgical Scene UnderstandingMultimodal LLMs / vision-language models → Medical vision-language models
Masks Can Be Distracting: On Context Comprehension in Diffusion Language ModelsDiffusion models → Discrete and text diffusion
TSGM: Regular and Irregular Time-series Generation using Score-based Generative ModelsDiffusion models → Diffusion foundations and sampling
Sawtooth Sampling for Time Series Denoising Diffusion Implicit ModelsDiffusion models → Diffusion foundations and sampling
PathMamba: A Hybrid Mamba-Transformer for Topologically Coherent Road Segmentation in Satellite ImageryState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
Co-Training Vision Language Models for Remote Sensing Multi-task LearningMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Multi-Reward GRPO for Stable and Prosodic Single-Codebook TTS LLMs at ScaleLLM reasoning → Reinforcement learning for reasoning
LaGen: Towards Autoregressive LiDAR Scene GenerationWorld models → World modeling from LiDAR generation
AuthenLoRA: Entangling Stylization with Imperceptible Watermarks for Copyright-Secure LoRA AdaptersDiffusion models → Robustness, privacy and adversarial aspects
From Diffusion to One-Step Generation: A Comparative Study of Flow-Based Models with Application to Image InpaintingDiffusion models → 3D, point cloud and scene generation
Self-Guided Adaptive Safety Alignment: Synthesizing and Internalizing Guidelines in Reasoning ModelsLLM reasoning → Reasoning for agents, planning and code +2
When Robots Obey the Patch: Universal Transferable Patch Attacks on Vision-Language-Action ModelsVision-language-action models → Safety and adversarial attacks
Scenes as Tokens: Multi-Scale Normal Distributions Transform Tokenizer for General 3D Vision-Language UnderstandingMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
Progress by Pieces: Test-Time Scaling for Autoregressive Image GenerationLLM reasoning → Multimodal reasoning
CAHS-Attack: CLIP-Aware Heuristic Search Attack Method for Stable DiffusionDiffusion models → Robustness, privacy and adversarial aspects +1
Kinematics-Aware Multi-Policy Reinforcement Learning for Force-Capable Humanoid Loco-ManipulationDeep reinforcement learning methods → RL for robot locomotion and manipulation
LLaVA-UHD v3: Progressive Visual Compression for Efficient Native-Resolution Encoding in MLLMsMultimodal LLMs / vision-language models → Visual token pruning and efficiency
Maglev-Pentabot: Magnetic Levitation System for Non-Contact Manipulation using Deep Reinforcement LearningDeep reinforcement learning methods → RL for robot locomotion and manipulation
TEAR: Temporal-aware Automated Red-teaming for Text-to-Video ModelsLLM safety and security → Vision-language and text-to-image safety
Efficient Training for Human Video Generation with Entropy-Guided Prioritized Progressive LearningDiffusion models → Video generation
CtrlVDiff: Controllable Video Generation via Unified Multimodal Video DiffusionDiffusion models → Video generation
Beyond Patch Aggregation: 3-Pass Pyramid Indexing for Vision-Enhanced Document RetrievalRetrieval-augmented generation → Multimodal RAG

All 378