Cross-Modal Latent Influence: Patch-Level Training Sample Attribution for Vision-Language Models
cs.AI
Submitted: 2026-10-09
Updated: 2026-10-09
Terminology
Sources
- Sparse Autoencoders Find Highly Interpretable Features in Language Models
- STRIDE: Training Data Attribution via Sparse Recovery from Subset Perturbations
- From What to How: Attributing CLIP's Latent Components Reveals Unexpected Semantic Reliance
- Scaling and evaluating sparse autoencoders
- Does Object Grounding Really Reduce Hallucination of Large Vision-Language Models?
- Studying Large Language Model Generalization with Influence Functions
- GQA: A New Dataset for Real-World Visual Reasoning and Compositional Question Answering
- Concept Influence: Leveraging Interpretability to Improve Performance and Efficiency in Training Data Attribution
- Extraction and Analysis of Multimodal Concepts in Vision Language Models through Sparse Autoencoders
- When Structured Sparse Autoencoders Learn Consistent Concepts Across Modalities
- SAE-V: Interpreting Multimodal Models for Enhanced Alignment
- SPARC: Concept-Aligned Sparse Autoencoders for Cross-Model and Cross-Modal Interpretability
- Sparse Autoencoders Learn Monosemantic Features in Vision-Language Models
- Do Vision & Language Decoders use Images and Text equally? How Self-consistent are their Explanations?
- SAVE: Sparse Autoencoder-Driven Visual Information Enhancement for Mitigating Object Hallucination
- Concept-TRAK: Understanding how diffusion models learn concepts through concept-level attribution
- VL-SAE: Interpreting and Enhancing Vision-Language Alignment with a Unified Concept Set
- Steering Vision-Language Models with Joint Sparse Autoencoders
- Gemma 4 Technical Report
- Qwen3 Technical Report
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems