Papers

New on arXiv

Thu 6 Feb 2025

351 papers

DistrEE: Distributed Early Exit of Deep Neural Network Inference on Edge DevicesEfficient LLM inference and serving → Early exit and edge inference
Vision-Integrated LLMs for Autonomous Driving Assistance : Human Performance Comparison and Trust EvaluationEnd-to-end autonomous driving → Large language models for driving
Agentic AI for Scaling Diagnosis and Care in Neurodegenerative DiseaseLLM agents and tool use → Medical and scientific domain agents
Reinforcement Learning on Dyads to Enhance Medication AdherenceDeep reinforcement learning methods → RL for clinical treatment and surgery
DiffNMR2: NMR Guided Sampling Acquisition Through Diffusion Model UncertaintyDiffusion models → Flow matching, time series and science
Robotouille: An Asynchronous Planning Benchmark for LLM AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
A Survey on Backdoor Threats in Large Language Models (LLMs): Attacks, Defenses, and EvaluationsLLM safety and security → Backdoor and data poisoning attacks
Preference Optimization via Contrastive Divergence: Your Reward Model is Secretly an NLL EstimatorLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
WaferLLM: Large Language Model Inference at Wafer ScaleEfficient LLM inference and serving → Serving systems and request scheduling
Mixture of neural operator experts for learning boundary conditions and model selectionMixture-of-experts → MoE theory and gating functions
Probing a Vision-Language-Action Model for Symbolic States and Integration into a Cognitive ArchitectureVision-language-action models → Reasoning and chain-of-thought in VLA
Speeding up Speculative Decoding via Sequential Approximate VerificationEfficient LLM inference and serving → Speculative decoding
Reinforcement Learning Based Prediction of PID Controller Gains for Quadrotor UAVsDeep reinforcement learning methods → RL for robot navigation and UAV control
Mechanisms of Projective Composition of Diffusion ModelsDiffusion models → Diffusion foundations and sampling
Multilingual Non-Autoregressive Machine Translation without Knowledge DistillationKnowledge distillation → Knowledge distillation for LLMs
Towards Cost-Effective Reward Guided Text GenerationLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF +1
Safety Must Precede the Deployment of Open-Ended AIDeep reinforcement learning methods → Safe and constrained RL
Beyond Sample-Level Feedback: Using Reference-Level Feedback to Guide Data SynthesisSynthetic data and self-improvement for LLMs → Code and instruction data synthesis
Fast Video Generation with Sliding Tile AttentionDiffusion models → Video generation
Revisiting Intermediate-Layer Matching in Knowledge Distillation: Layer-Selection Strategy Doesn't Matter (Much)Knowledge distillation → General knowledge distillation methods
Dynamic Optimizations of LLM Ensembles with Two-Stage Reinforcement Learning AgentsLLM agents and tool use → Reinforcement learning for LLM agents
Provable Sample-Efficient Transfer Learning Conditional Diffusion Models via Representation LearningDiffusion models → Diffusion foundations and sampling
Active Task Disambiguation with LLMsLLM agents and tool use → General LLM agent frameworks and benchmarks
Augmented Conditioning Is Enough For Effective Training Image GenerationDiffusion models → Image generation, text-to-image and editing
Color in Visual-Language Models: CLIP deficienciesMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Iterative Importance Fine-tuning of Diffusion ModelsDiffusion models → Diffusion foundations and sampling
Training Language Models to Reason EfficientlyLLM reasoning → Reinforcement learning for reasoning
EmoBench-M: Benchmarking Emotional Intelligence for Multimodal Large Language ModelsMultimodal LLMs / vision-language models → Multimodal emotion understanding
KVTuner: Sensitivity-Aware Layer-Wise Mixed-Precision KV Cache Quantization for Efficient and Nearly Lossless LLM InferenceEfficient LLM inference and serving → KV cache quantization
Understanding and Mitigating Bias Inheritance in LLM-based Data Augmentation on Downstream TasksSynthetic data and self-improvement for LLMs → LLM synthetic data generation
Autotelic Reinforcement Learning: Exploring Intrinsic Motivations for Skill Acquisition in Open-Ended EnvironmentsDeep reinforcement learning methods → Deep RL algorithms and representations
Analytical FFN-to-MoE Restructuring via Activation Pattern AnalysisMixture-of-experts → MoE large language models
MedRAG: Enhancing Retrieval-augmented Generation with Knowledge Graph-Elicited Reasoning for Healthcare CopilotRetrieval-augmented generation → Medical and biomedical RAG
Decoder-Only LLMs are Better Controllers for Diffusion ModelsDiffusion models → Image generation, text-to-image and editing
Step Back to Leap Forward: Self-Backtracking for Boosting Reasoning of Language ModelsLLM reasoning → Test-time compute and efficient thinking
Agency Is Frame-DependentDeep reinforcement learning methods → Deep RL algorithms and representations
Time-VLM: Exploring Multimodal Vision-Language Models for Augmented Time Series ForecastingMultimodal LLMs / vision-language models → Chart understanding and anomaly detection
UniCP: A Unified Caching and Pruning Framework for Efficient Video GenerationDiffusion models → Video generation
Division-of-Thoughts: Harnessing Hybrid Language Model Synergy for Efficient On-Device AgentsLLM reasoning → Reasoning for agents, planning and code +1
Value-Based Deep RL Scales PredictablyDeep reinforcement learning methods → Deep RL algorithms and representations
WorldSense: Evaluating Real-world Omnimodal Understanding for Multimodal LLMsMultimodal LLMs / vision-language models → Video understanding with MLLMs
Speak Easy: Eliciting Harmful Jailbreaks from LLMs with Simple InteractionsLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
ConceptAttention: Diffusion Transformers Learn Highly Interpretable FeaturesDiffusion models → Image generation, text-to-image and editing
Great Models Think Alike and this Undermines AI OversightLLM reasoning → Reasoning benchmarks and evaluation
HOG-Diff: Higher-Order Guided Diffusion for Graph GenerationDiffusion models → Flow matching, time series and science
ScoreFlow: Mastering LLM Agent Workflows via Score-based Preference OptimizationLLM agents and tool use → Software engineering coding agents +1
MotionCanvas: Cinematic Shot Design with Controllable Image-to-Video GenerationDiffusion models → Video generation
PILAF: Optimal Human Preference Sampling for Reward ModelingLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
Cross the Gap: Exposing the Intra-modal Misalignment in CLIP via Modality InversionMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Realistic Image-to-Image Machine Unlearning via Decoupling and Knowledge RetentionLLM safety and security → Vision-language and text-to-image safety
Cognitive AI framework 2.0: advances in the simulation of human thoughtLLM reasoning → Reasoning benchmarks and evaluation
Free Energy Risk Metrics for Systemically Safe AI: Gatekeeping Multi-Agent StudyWorld models → Latent dynamics models and model-based RL +1
Short-length Adversarial Training Helps LLMs Defend Long-length Jailbreak Attacks: Theoretical and Empirical EvidenceLLM safety and security → Jailbreaking
PixFoundation: Are We Heading in the Right Direction with Pixel-level Vision Foundation Models?Multimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Multi-agent Architecture Search via Agentic SupernetLLM agents and tool use → General LLM agent frameworks and benchmarks
MRAMG-Bench: A Comprehensive Benchmark for Advancing Multimodal Retrieval-Augmented Multimodal GenerationRetrieval-augmented generation → Multimodal RAG
Behavioral Entropy-Guided Dataset Generation for Offline Reinforcement LearningOffline RL and imitation learning from data → Offline RL algorithms
Llasa: Scaling Train-Time and Inference-Time Compute for Llama-based Speech SynthesisLLM reasoning → Audio-language model reasoning
VTutor: An Open-Source SDK for Generative AI-Powered Animated Pedagogical Agents with Multi-Media OutputLLM agents and tool use → Educational agents and math problem solving
Efficient Few-Shot Continual Learning in Vision-Language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
AttentionPredictor: Temporal Patterns Matter for KV Cache CompressionEfficient LLM inference and serving → KV cache optimization
TQ-DiT: Efficient Time-Aware Quantization for Diffusion TransformersDiffusion models → Diffusion foundations and sampling
PartEdit: Fine-Grained Image Editing using Pre-Trained Diffusion ModelsDiffusion models → Image generation, text-to-image and editing
Safety Reasoning with GuidelinesLLM reasoning → Reasoning for agents, planning and code
Fine, I'll Merge It Myself: A Multi-Fidelity Framework for Automated Model MergingLLM reasoning → Test-time compute and efficient thinking
CAD-Editor: A Locate-then-Infill Framework with Automated Training Data Synthesis for Text-Based CAD EditingSynthetic data and self-improvement for LLMs → Self-improving MLLMs and synthetic data
RWKV-UI: UI Understanding with Enhanced Perception and ReasoningLLM reasoning → Multimodal reasoning +1
Bilevel Multi-Armed Bandit-Based Hierarchical Reinforcement Learning for Interaction-Aware Self-Driving at Unsignalized IntersectionsEnd-to-end autonomous driving → Reinforcement learning for driving +1
Fairness Aware Reinforcement Learning via Proximal Policy OptimizationDeep reinforcement learning methods → RL for games and combinatorial optimization
Enhancing Online Learning Efficiency Through Heterogeneous Resource Integration with a Multi-Agent RAG SystemRetrieval-augmented generation → Agentic RAG, agents and memory
CleanSurvival: Automated data preprocessing for time-to-event models using reinforcement learningDeep reinforcement learning methods → Deep RL algorithms and representations
DiTAR: Diffusion Transformer Autoregressive Modeling for Speech GenerationDiffusion models → Audio, speech and music
Experiments with Large Language Models on Retrieval-Augmented Generation for Closed-Source Simulation SoftwareRetrieval-augmented generation → RAG pipeline design and evaluation
UniForm: A Unified Multi-Task Diffusion Transformer for Audio-Video GenerationDiffusion models → Audio, speech and music
Rank Also Matters: Hierarchical Configuration for Mixture of Adapter Experts in LLM Fine-TuningMixture-of-experts → Mixture of LoRA experts
BOLT: Bootstrap Long Chain-of-Thought in Language Models without DistillationLLM reasoning → Prompting and chain-of-thought reasoning +1
Mirror Descent Actor Critic via Bounded Advantage LearningDeep reinforcement learning methods → RL theory and sample complexity
Improving Natural Language Understanding for LLMs via Large-Scale Instruction SynthesisLLM alignment / RLHF / preference optimisation → LLM preference optimization methods +1
Syntriever: How to Train Your Retriever with Synthetic Data from LLMsSynthetic data and self-improvement for LLMs → LLM synthetic data generation
Dynamic Rank Adjustment in Diffusion Policies for Efficient and Flexible TrainingDiffusion models → Diffusion and flow policies for robotics +1
PsyPlay: Personality-Infused Role-Playing Conversational AgentsLLM agents and tool use → Social simulation and multi-agent debate
DeblurDiff: Real-World Image Deblurring with Generative Diffusion ModelsDiffusion models → Inverse problems and image restoration
CriticalKV: Optimizing KV Cache Eviction from an Output Perturbation PerspectiveEfficient LLM inference and serving → KV cache optimization
SoK: Benchmarking Poisoning Attacks and Defenses in Federated LearningLLM safety and security → Federated unlearning and poisoning
Frontend Diffusion: Empowering Self-Representation of Junior Researchers and Designers Through Multi-agent SystemLLM reasoning → Reasoning for agents, planning and code
Iterate to Accelerate: A Unified Framework for Iterative Reasoning and Feedback ConvergenceLLM reasoning → Test-time compute and efficient thinking
A Retrospective Systematic Study on Hierarchical Sparse Query Transformer-assisted Ultrasound Screening for Early Hepatocellular CarcinomaMixture-of-experts → Medical and biomedical MoE
Learning Reward Machines from Partially Observed PoliciesOffline RL and imitation learning from data → Inverse reinforcement learning +1
Self-Improving Skill Learning for Robust Skill-based Meta-Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Rethinking Residual Distribution in Locate-then-Edit Model EditingLLM reasoning → Knowledge- and tool-augmented reasoning
Action-Free Reasoning for Policy GeneralizationOffline RL and imitation learning from data → Robot manipulation imitation learning
DICE: Distilling Classifier-Free Guidance into Text EmbeddingsDiffusion models → Diffusion foundations and sampling
Seeing in the Dark: A Teacher-Student Framework for Dark Video Action Recognition via Knowledge Distillation and Contrastive LearningKnowledge distillation → General knowledge distillation methods
Speaking the Language of Teamwork: LLM-Guided Credit Assignment in Multi-Agent Reinforcement LearningLLM agents and tool use → Reinforcement learning for LLM agents
Universal Sparse Autoencoders: Interpretable Cross-Model Concept AlignmentMechanistic interpretability → Interpreting vision and multimodal models
LLM Alignment as Retriever Optimization: An Information Retrieval PerspectiveLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
How Vulnerable Is My Learned Policy? Universal Adversarial Perturbation Attacks On Modern Behavior Cloning PoliciesOffline RL and imitation learning from data → Imitation learning methods
DocMIA: Document-Level Membership Inference Attacks against DocVQA ModelsLLM safety and security → Membership inference attacks
Conditional Diffusion Models are Medical Image Classifiers that Provide Explainability and Uncertainty for FreeDiffusion models → Medical imaging and segmentation
Variational Control for Guidance in Diffusion ModelsDiffusion models → Diffusion foundations and sampling

All 351