Papers

New on arXiv

Wed 11 Dec 2024

352 papers

EvalSVA: Multi-Agent Evaluators for Next-Gen Software Vulnerability AssessmentLLM agents and tool use → Agent threats, guardrails and governance
A Multimodal Social AgentLLM agents and tool use → General LLM agent frameworks and benchmarks
Towards LLM-based optimization compilers. Can LLMs learn how to apply a single peephole optimization? Reasoning is all LLMs need!LLM reasoning → Test-time compute and efficient thinking
What Makes In-context Learning Effective for Mathematical Reasoning: A Theoretical AnalysisLLM reasoning → Reasoning benchmarks and evaluation
GraphTool-Instruction: Revolutionizing Graph Reasoning in LLMs through Decomposed Subtask InstructionLLM reasoning → Knowledge- and tool-augmented reasoning
SMARTCAL: An Approach to Self-Aware Tool-Use Evaluation and CalibrationLLM agents and tool use → Tool learning and function calling
Unlocking Visual Secrets: Inverting Features with Diffusion Priors for Image ReconstructionDiffusion models → Robustness, privacy and adversarial aspects
TidyBot++: An Open-Source Holonomic Mobile Manipulator for Robot LearningOffline RL and imitation learning from data → Robot manipulation imitation learning
SVGFusion: A VAE-Diffusion Transformer for Vector Graphic GenerationDiffusion models → Image generation, text-to-image and editing
COEF-VQ: Cost-Efficient Video Quality Understanding through a Cascaded Multimodal LLM FrameworkMultimodal LLMs / vision-language models → Image editing and quality assessment
NAT-NL2GQL: A Novel Multi-Agent Framework for Translating Natural Language to Graph Query LanguageLLM agents and tool use → Web, search, and research agents
Imitate Before Detect: Aligning Machine Stylistic Preference for Machine-Revised Text DetectionLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
DSplats: 3D Generation by Denoising Splats-Based Multiview Diffusion ModelsDiffusion models → 3D, point cloud and scene generation
Bench2Drive-R: Turning Real World Data into Reactive Closed-Loop Autonomous Driving Benchmark by Generative ModelEnd-to-end autonomous driving → Generative simulation for driving
HARP: A challenging human-annotated math reasoning benchmarkLLM reasoning → Mathematical reasoning
Test-Time Alignment via Hypothesis ReweightingLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
Generative Agents for Multi-Agent Autoformalization of Interaction ScenariosLLM agents and tool use → Reinforcement learning for LLM agents
Latent Safety-Constrained Policy Approach for Safe Offline Reinforcement LearningOffline RL and imitation learning from data → Safe offline RL
GMem: A Modular Approach for Ultra-Efficient Generative ModelsDiffusion models → Diffusion foundations and sampling
LLaVA-Zip: Adaptive Visual Token Compression with Intrinsic Image InformationMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
Proactive Adversarial Defense: Harnessing Prompt Tuning in Vision-Language Models to Detect Unseen Backdoored ImagesMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
DocVLM: Make Your VLM an Efficient ReaderMultimodal LLMs / vision-language models → Document understanding and multimodal RAG
Euclid: Supercharging Multimodal LLMs with Synthetic High-Fidelity Visual DescriptionsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
A quantum-classical reinforcement learning model to play Atari gamesDeep reinforcement learning methods → Quantum reinforcement learning
VisionArena: 230K Real World User-VLM Conversations with Preference LabelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Efficient Gravitational Wave Parameter Estimation via Knowledge Distillation: A ResNet1D-IAF ApproachKnowledge distillation → General knowledge distillation methods
StreamChat: Chatting with Streaming VideoMultimodal LLMs / vision-language models → Video understanding with MLLMs
GPD-1: Generative Pre-training for DrivingEnd-to-end autonomous driving → Generative simulation for driving
DMin: Scalable Training Data Influence Estimation for Diffusion ModelsDiffusion models → Diffusion foundations and sampling
Multimodal Latent Language Modeling with Next-Token DiffusionMultimodal LLMs / vision-language models → Image editing and quality assessment
Physics Context Builders: A Modular Framework for Physical Reasoning in Vision-Language ModelsLLM reasoning → Spatial and embodied reasoning +1
Exploiting the Index Gradients for Optimization-Based Jailbreaking on Large Language ModelsLLM safety and security → Jailbreaking
Benchmarking Large Vision-Language Models via Directed Scene Graph for Comprehensive Image CaptioningMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
AdvWave: Stealthy Adversarial Jailbreak Attack against Large Audio-Language ModelsLLM safety and security → Audio jailbreaks and speech backdoors
Leveraging Graph-RAG and Prompt Engineering to Enhance LLM-Based Automated Requirement Traceability and Compliance ChecksRetrieval-augmented generation → RAG pipeline design and evaluation
TurboAttention: Efficient Attention Approximation For High Throughputs LLMsEfficient LLM inference and serving → KV cache optimization
Sketch Decompositions for Classical Planning via Deep Reinforcement LearningDeep reinforcement learning methods → Deep RL algorithms and representations
TryOffAnyone: Tiled Cloth Generation from a Dressed PersonDiffusion models → 3D, point cloud and scene generation
An End-to-End Collaborative Learning Approach for Connected Autonomous Vehicles in Occluded ScenariosEnd-to-end autonomous driving → Cooperative and V2X driving +1
Underestimated Privacy Risks for Minority Populations in Large Language Model UnlearningLLM safety and security → Machine unlearning
Watermarking Training Data of Music Generation ModelsLLM safety and security → Watermarking generated content and models
SenCLIP: Enhancing zero-shot land-use mapping for Sentinel-2 with ground-level promptingMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
EMS: Adaptive Evict-then-Merge Strategy for Head-wise KV Cache Compression Based on Global-Local ImportanceEfficient LLM inference and serving → KV cache optimization
Bridging Relevance and Reasoning: Rationale Distillation in Retrieval-Augmented GenerationLLM reasoning → Knowledge- and tool-augmented reasoning +1
SAM-Mamba: Mamba Guided SAM Architecture for Generalized Zero-Shot Polyp SegmentationState-space sequence models (S4, Mamba) → Mamba for medical image analysis
InvDiff: Invariant Guidance for Bias Mitigation in Diffusion ModelsDiffusion models → Diffusion foundations and sampling
Multi-perspective Alignment for Increasing Naturalness in Neural Machine TranslationLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Efficient Rectification of Neuro-Symbolic Reasoning Inconsistencies by Abductive ReflectionLLM reasoning → Reasoning benchmarks and evaluation
TapeAgents: a Holistic Framework for Agent Development and OptimizationLLM agents and tool use → General LLM agent frameworks and benchmarks
POINTS1.5: Building a Vision-Language Model towards Real World ApplicationsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
From Multimodal LLMs to Generalist Embodied Agents: Methods and LessonsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models +1
Pragmatist: Multiview Conditional Diffusion Models for High-Fidelity 3D Reconstruction from Unposed Sparse ViewsDiffusion models → 3D, point cloud and scene generation
Physical Informed Driving World ModelWorld models → World models for autonomous driving +1
Grasp Diffusion Network: Learning Grasp Generators from Partial Point Clouds with Diffusion Models in SO(3)xR3Diffusion models → Diffusion and flow policies for robotics
Adversarial Purification by Consistency-aware Latent Space Optimization on Data ManifoldsDiffusion models → Robustness, privacy and adversarial aspects
Learning to Reason via Self-Iterative Process Feedback for Small Language ModelsLLM reasoning → Reinforcement learning for reasoning +1
SweetieChat: A Strategy-Enhanced Role-playing Framework for Diverse Scenarios Handling Emotional Support AgentLLM agents and tool use → Social simulation and multi-agent debate
LOMA: Language-assisted Semantic Occupancy Network via Triplane MambaState-space sequence models (S4, Mamba) → Point clouds, detection and remote sensing
FILA: Fine-Grained Vision Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency
Video Summarization using Denoising Diffusion Probabilistic ModelDiffusion models → Video generation
SmolTulu: Higher Learning Rate to Batch Size Ratios Can Lead to Better Reasoning in SLMsLLM reasoning → Reasoning benchmarks and evaluation +1
CoDTS: Enhancing Sparsely Supervised Collaborative Perception with a Dual Teacher-Student FrameworkKnowledge distillation → General knowledge distillation methods
Large Language Models Still Face Challenges in Multi-Hop Reasoning with External KnowledgeLLM reasoning → Knowledge- and tool-augmented reasoning
Investigating the Scaling Effect of Instruction Templates for Training Multimodal Language ModelMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
SINERGYM -- A virtual testbed for building energy optimization with Reinforcement LearningDeep reinforcement learning methods → RL for energy, networks and operations
Self-Refining Diffusion Samplers: Enabling Parallelization via Parareal IterationsDiffusion models → Diffusion foundations and sampling
Collaborative Inference for Large Models with Task Offloading and Early ExitingEfficient LLM inference and serving → Early exit and edge inference
Lachesis: Predicting LLM Inference Accuracy using Structural Properties of Reasoning PathsLLM reasoning → Test-time compute and efficient thinking
FLIP: Flow-Centric Generative Planning as General-Purpose Manipulation World ModelWorld models → Embodied and robot world models
Generate Any Scene: Scene Graph Driven Data Synthesis for Visual Generation TrainingDiffusion models → Image generation, text-to-image and editing +1
Model-Editing-Based Jailbreak against Safety-aligned Large Language ModelsLLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks +1
Mixture of Experts Meets Decoupled Message Passing: Towards General and Adaptive Node ClassificationMixture-of-experts → MoE for time series and graphs
RAG-IT: Retrieval-Augmented Instruction Tuning for Automated Financial Analysis -- A Case Study for the Semiconductor SectorRetrieval-augmented generation → Legal and financial RAG
TextRefiner: Internal Visual Feature as Efficient Refiner for Vision-Language Models Prompt TuningMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Analyzing and Mitigating Model Collapse in Rectified Flow ModelsDiffusion models → Diffusion foundations and sampling +1
DG-Mamba: Robust and Efficient Dynamic Graph Structure Learning with Selective State Space ModelsState-space sequence models (S4, Mamba) → Video, motion and generative Mamba models
How Vision-Language Tasks Benefit from Large Pre-trained Models: A SurveyLLM reasoning → Multimodal reasoning
Antelope: Potent and Concealed Jailbreak Attack StrategyLLM safety and security → Jailbreaking
Wasserstein Distance Rivals Kullback-Leibler Divergence for Knowledge DistillationKnowledge distillation → General knowledge distillation methods
DiffRaman: A Conditional Latent Denoising Diffusion Probabilistic Model for Bacterial Raman Spectroscopy Identification Under Limited Data ConditionsDiffusion models → Flow matching, time series and science
Progressive Multi-granular Alignments for Grounded Reasoning in Large Vision-Language ModelsLLM reasoning → Multimodal reasoning +1
LatentSpeech: Latent Diffusion for Text-To-Speech GenerationDiffusion models → Audio, speech and music
Diffusion-based Data Augmentation and Knowledge Distillation with Generated Soft Labels Solving Data Scarcity Problems of SAR Oil Spill SegmentationDiffusion models → Inverse problems and image restoration
Seeing Syntax: Uncovering Syntactic Learning Limitations in Vision-Language ModelsMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
The ART of Composition: Attention-Regularized Training for Compositional Visual GroundingMultimodal LLMs / vision-language models → VLM benchmarks and visual reasoning
Exploiting Vision Encoder Vulnerabilities for Universal Adversarial Perturbations on Large Vision-Language ModelsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
Multilingual LLMs Inherently Reward In-Language Time-Sensitive Semantic Alignment for Low-Resource LanguagesLLM reasoning → Reasoning benchmarks and evaluation
DialogAgent: An Auto-engagement Agent for Code Question Answering Data ProductionLLM agents and tool use → Software engineering coding agents
Federated In-Context LLM Agent LearningLLM reasoning → Reasoning for agents, planning and code +1
EmoVerse: Exploring Multimodal Large Language Models for Sentiment and Emotion UnderstandingMultimodal LLMs / vision-language models → Multimodal emotion understanding
Bootstrapping Heterogeneous Graph Representation Learning via Large Language Models: A Generalized ApproachSynthetic data and self-improvement for LLMs → Synthetic reasoning data and bootstrapping
TinyThinker: Distilling Reasoning through Coarse-to-Fine Knowledge Internalization with Self-ReflectionLLM reasoning → Test-time compute and efficient thinking +1
Can a MISL Fly? Analysis and Ingredients for Mutual Information Skill LearningDeep reinforcement learning methods → Deep RL algorithms and representations
Ask1: Development and Reinforcement Learning-Based Control of a Custom Quadruped RobotDeep reinforcement learning methods → RL for robot locomotion and manipulation
MAGIC: Mastering Physical Adversarial Generation in Context through Collaborative LLM AgentsLLM agents and tool use → Agent threats, guardrails and governance
Adaptive Querying for Reward Learning from Human FeedbackLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
NEAT Algorithm-based Stock Trading Strategy with Multiple Technical Indicators Resonance
Navigating Ethical Challenges in Generative AI-Enhanced Research: The ETHICAL Framework for Responsible Generative AI Use
Enhancing Drug-Target Interaction Prediction through Transfer Learning from Activity Cliff Prediction Tasks
Predicting Human Brain States with Transformer

All 352