Papers

New on arXiv

Thu 20 Feb 2025

414 papers

Verify when Uncertain: Beyond Self-Consistency in Black Box Hallucination DetectionLLM reasoning → Hallucination detection and mitigation
Vending-Bench: A Benchmark for Long-Term Coherence of Autonomous AgentsLLM agents and tool use → General LLM agent frameworks and benchmarks
Soft Token Attacks Cannot Reliably Audit Unlearning in Large Language ModelsLLM safety and security → Machine unlearning
Pragmatic Reasoning improves LLM Code GenerationLLM reasoning → Reasoning for agents, planning and code
Show Me Your Code! Kill Code Poisoning: A Lightweight Method Based on Code NaturalnessLLM safety and security → Backdoor and data poisoning attacks
A Stronger Mixture of Low-Rank Experts for Fine-Tuning Foundation ModelsMixture-of-experts → Mixture of LoRA experts
CoME: An Unlearning-based Approach to Conflict-free Model EditingLLM safety and security → Machine unlearning
Universal AI maximizes Variational EmpowermentDeep reinforcement learning methods → Deep RL algorithms and representations
LUME: LLM Unlearning with Multitask EvaluationsLLM safety and security → Machine unlearning
ExpertLens: Activation steering features are highly interpretableMechanistic interpretability → Activation steering methods
Is Safety Standard Same for Everyone? User-Specific Safety Evaluation of Large Language ModelsLLM safety and security → Safety alignment, refusal and guardrails
UPCORE: Utility-Preserving Coreset Selection for Balanced UnlearningLLM safety and security → Machine unlearning
Can Hallucination Correction Improve Video-Language Alignment?Multimodal LLMs / vision-language models → Hallucination mitigation in LVLMs
Hardware-Friendly Static Quantization Method for Video Diffusion TransformersDiffusion models → Diffusion foundations and sampling
Quantize What Counts: More for Keys, Less for ValuesEfficient LLM inference and serving → KV cache quantization
Pseudoinverse Diffusion Models for Generative CT Image Reconstruction from Low Dose DataDiffusion models → Medical imaging and segmentation
DDAT: Diffusion Policies Enforcing Dynamically Admissible Robot TrajectoriesDiffusion models → Diffusion and flow policies for robotics
Reducing Hallucinations of Medical Multimodal Large Language Models with Visual Retrieval-Augmented GenerationMultimodal LLMs / vision-language models → Hallucination mitigation in LVLMs +1
InterFeedback: Unveiling Interactive Intelligence of Large Multimodal Models via Human FeedbackMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning +1
Is Relevance Propagated from Retriever to Generator in RAG?Retrieval-augmented generation → RAG pipeline design and evaluation
Obliviate: Efficient Unmemorization for Protecting Intellectual Property in Large Language ModelsLLM safety and security → Machine unlearning
A Socratic RAG Approach to Connect Natural Language Queries on Research Topics with Knowledge Organization SystemsRetrieval-augmented generation → Agentic RAG, agents and memory
EigenShield: Causal Subspace Filtering via Random Matrix Theory for Adversarially Robust Vision-Language ModelsMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness
Beyond No: Quantifying AI Over-Refusal and Emotional Attachment BoundariesLLM safety and security → Safety alignment, refusal and guardrails
Learning to Solve and Verify: A Self-Play Framework for Code and Test GenerationSynthetic data and self-improvement for LLMs → Code and instruction data synthesis
Reward-Guided Iterative Refinement in Diffusion Models at Test-Time with Applications to Protein and DNA DesignDiffusion models → Flow matching, time series and science
FacaDiffy: Inpainting Unseen Facade Parts Using Diffusion ModelsDiffusion models → 3D, point cloud and scene generation
Learning to Retrieve and Reason on Knowledge Graph through Active Self-ReflectionLLM reasoning → Knowledge- and tool-augmented reasoning
RAGVA: Engineering Retrieval Augmented Generation-based Virtual Assistants in PracticeRetrieval-augmented generation → RAG pipeline design and evaluation
LServe: Efficient Long-sequence LLM Serving with Unified Sparse AttentionEfficient LLM inference and serving → KV cache optimization
Benchmarking Multimodal RAG through a Chart-based Document Question-Answering Generation FrameworkLLM reasoning → Multimodal reasoning +1
ALFA: Aligning LLMs to Ask Good Questions A Case Study in Clinical ReasoningLLM reasoning → Clinical and biomedical applications +1
FR-Spec: Accelerating Large-Vocabulary Language Models via Frequency-Ranked Speculative SamplingEfficient LLM inference and serving → Speculative decoding
CLIPPER: Compression enables long-context synthetic data generationSynthetic data and self-improvement for LLMs → LLM synthetic data generation
Red-Teaming LLM Multi-Agent Systems via Communication AttacksLLM agents and tool use → Agent threats, guardrails and governance +1
Scaling Text-Rich Image Understanding via Code-Guided Synthetic Multimodal Data GenerationMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMsLLM reasoning → Test-time compute and efficient thinking +1
LongWriter-V: Enabling Ultra-Long and High-Fidelity Generation in Vision-Language ModelsMultimodal LLMs / vision-language models → Video understanding with MLLMs
Improving the Diffusability of AutoencodersDiffusion models → Diffusion foundations and sampling
Measuring Chain of Thought Faithfulness by Unlearning Reasoning StepsLLM reasoning → Prompting and chain-of-thought reasoning +1
From RAG to Memory: Non-Parametric Continual Learning for Large Language ModelsRetrieval-augmented generation → Generation and context integration
A Multi-Agent Perspective on Modern Information RetrievalLLM agents and tool use → Web, search, and research agents
Ray-Tracing for Conditionally Activated Neural NetworksMixture-of-experts → MoE theory and gating functions
DC-ControlNet: Decoupling Inter- and Intra-Element Conditions in Image Generation with Diffusion ModelsDiffusion models → Image generation, text-to-image and editing
Harnessing PDF Data for Improving Japanese Large Multimodal ModelsMultimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement LearningLLM reasoning → Reinforcement learning for reasoning
Tree-of-Debate: Multi-Persona Debate Trees Elicit Critical Thinking for Scientific Comparative AnalysisLLM reasoning → Reasoning for agents, planning and code
Step-by-Step Fact Verification System for Medical Claims with Explainable ReasoningLLM reasoning → Reasoning benchmarks and evaluation
On the Influence of Context Size and Model Choice in Retrieval-Augmented Generation SystemsRetrieval-augmented generation → RAG robustness and adaptive retrieval
HiddenDetect: Detecting Jailbreak Attacks against Large Vision-Language Models via Monitoring Hidden StatesMultimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness +1
Multi-Agent Coordination across Diverse Applications: A SurveyLLM agents and tool use → General LLM agent frameworks and benchmarks
Reinforcement Learning with Graph Attention for Routing and Wavelength Assignment with Lightpath ReuseDeep reinforcement learning methods → RL for energy, networks and operations
WavRAG: Audio-Integrated Retrieval Augmented Generation for Spoken Dialogue ModelsRetrieval-augmented generation → Conversational and speech RAG
I-MCTS: Enhancing Agentic AutoML via Introspective Monte Carlo Tree SearchLLM agents and tool use → General LLM agent frameworks and benchmarks
AlphaMaze: Enhancing Large Language Models' Spatial Intelligence via GRPOLLM reasoning → Spatial and embodied reasoning
iAgent: LLM Agent as a Shield between User and Recommender SystemsLLM agents and tool use → Agentic recommender systems
Length-Controlled Margin-Based Preference Optimization without Reference ModelLLM reasoning → Reinforcement learning for reasoning +1
NAVIG: Natural Language-guided Analysis with Vision Language Models for Image Geo-localizationMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
CER: Confidence Enhanced Reasoning in LLMsLLM reasoning → Test-time compute and efficient thinking
Reward Models Identify Consistency, Not CausalityLLM alignment / RLHF / preference optimisation → Reward modeling and RLHF
SageServe: Optimizing LLM Serving on Cloud Data Centers with Forecast Aware Auto-ScalingEfficient LLM inference and serving → Serving systems and request scheduling
ICA-RAG: Information Completeness Guided Adaptive Retrieval-Augmented Generation for Disease DiagnosisRetrieval-augmented generation → Medical and biomedical RAG
Noisy Test-Time Adaptation in Vision-Language ModelsMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
ReVISE: Learning to Refine at Test-Time via Intrinsic Self-VerificationLLM reasoning → Test-time compute and efficient thinking
Plan-over-Graph: Towards Parallelable LLM Agent ScheduleLLM agents and tool use → General LLM agent frameworks and benchmarks
Less is More: Improving LLM Alignment via Preference Data SelectionLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
Model Inversion Attack against Federated UnlearningLLM safety and security → Federated unlearning and poisoning
CORBA: Contagious Recursive Blocking Attacks on Multi-Agent Systems Based on Large Language ModelsLLM agents and tool use → Agent threats, guardrails and governance
PLPHP: Per-Layer Per-Head Vision Token Pruning for Efficient Large Vision-Language ModelsMultimodal LLMs / vision-language models → Visual token pruning and efficiency +1
MLGym: A New Framework and Benchmark for Advancing AI Research AgentsLLM agents and tool use → Web, search, and research agents
Advancing Language Multi-Agent Learning with Credit Re-Assignment for Interactive Environment GeneralizationLLM agents and tool use → Reinforcement learning for LLM agents
How Jailbreak Defenses Work and Ensemble? A Mechanistic InvestigationLLM safety and security → Jailbreaking
Llamba: Scaling Distilled Recurrent Models for Efficient Language ProcessingState-space sequence models (S4, Mamba) → Core SSM architectures for language
Early-Exit and Instant Confidence Translation Quality EstimationEfficient LLM inference and serving → Early exit and edge inference
ChatVLA: Unified Multimodal Understanding and Robot Control with Vision-Language-Action ModelVision-language-action models → Reasoning and chain-of-thought in VLA
Evaluating Precise Geolocation Inference Capabilities of Vision Language ModelsMultimodal LLMs / vision-language models → Embodied and 3D vision-language models
HPS: Hard Preference Sampling for Human Preference AlignmentLLM alignment / RLHF / preference optimisation → LLM preference optimization methods +1
S*: Test Time Scaling for Code GenerationLLM reasoning → Reasoning for agents, planning and code
RelaCtrl: Relevance-Guided Efficient Control for Diffusion TransformersDiffusion models → Diffusion foundations and sampling
A Similarity Paradigm Through Textual Regularization Without ForgettingMultimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP
Discovering highly efficient low-weight quantum error-correcting codes with reinforcement learningDeep reinforcement learning methods → Quantum reinforcement learning
PPO-MI: Efficient Black-Box Model Inversion via Proximal Policy OptimizationDeep reinforcement learning methods → RL for energy, networks and operations
Is Q-learning an Ill-posed Problem?Deep reinforcement learning methods → RL theory and sample complexity
Topology-Aware Wavelet Mamba for Airway Structure Segmentation in Postoperative Recurrent Nasopharyngeal Carcinoma CT ScansState-space sequence models (S4, Mamba) → Mamba for medical image analysis
Retrieval-Augmented Process Reward Model for Generalizable Mathematical ReasoningLLM reasoning → Reinforcement learning for reasoning +2
Full-Step-DPO: Self-Supervised Preference Optimization with Step-wise Rewards for Mathematical ReasoningLLM reasoning → Mathematical reasoning +1
Self-Improvement Towards Pareto Optimality: Mitigating Preference Conflicts in Multi-Objective AlignmentLLM reasoning → Reinforcement learning for reasoning +2
SR-LLM: Rethinking the Structured Representation in Large Language ModelLLM reasoning → Reasoning benchmarks and evaluation
FlowAgent: Achieving Compliance and Flexibility for Workflow AgentsLLM agents and tool use → Software engineering coding agents
Earlier Tokens Contribute More: Learning Direct Preference Optimization From Temporal Decay PerspectiveLLM alignment / RLHF / preference optimisation → LLM preference optimization methods
A Survey on Feedback-based Multi-step Reasoning for Large Language Models on MathematicsLLM reasoning → Mathematical reasoning
Beyond Self-Talk: A Communication-Centric Survey of LLM-Based Multi-Agent SystemsLLM agents and tool use → General LLM agent frameworks and benchmarks
ParallelComp: Parallel Long-Context Compressor for Length ExtrapolationEfficient LLM inference and serving → KV cache optimization
Textured 3D Regenerative Morphing with 3D Diffusion PriorDiffusion models → 3D, point cloud and scene generation
μRL: Discovering Transient Execution Vulnerabilities Using Reinforcement LearningDeep reinforcement learning methods → RL training of language and agent models
Drift: Decoding-time Personalized Alignments with Implicit User PreferencesLLM alignment / RLHF / preference optimisation → Human values and AI feedback alignment
PC-Agent: A Hierarchical Multi-Agent Collaboration Framework for Complex Task Automation on PCLLM agents and tool use → GUI and mobile agents
STeCa: Step-level Trajectory Calibration for LLM Agent LearningLLM reasoning → Reasoning for agents, planning and code +1
Capturing Nuanced Preferences: Preference-Aligned Distillation for Small Language ModelsLLM reasoning → Reinforcement learning for reasoning +2
PaperHelper: Knowledge-Based LLM QA Paper Reading AssistantRetrieval-augmented generation → RAG pipeline design and evaluation

All 414