DocShield: AI-Powered Document Safety with Evidence-Based Reasoning
Discover DocShield, an AI framework enhancing document safety via visual-logical reasoning and evidence-grounded forgery detection for secure digital docum...
Efficient3D: Adaptive Token Reduction for 3D MLLMs
Discover Efficient3D, a framework that boosts 3D MLLMs efficiency with adaptive, debiased token reduction, improving inference speed and accuracy.
Token-Space Attacks on Reward Models in RLHF
Discover how token-space attacks exploit reward models in RLHF, revealing vulnerabilities beyond semantic manipulation and impacting AI safety.
Discovering Belief Geometries Using Sparse Autoencoders
Explore how sparse autoencoders reveal belief-state geometries in transformer models, advancing mechanistic interpretability of large language models.
EligMeta: Eligibility-Aware Framework for Clinical Trial Meta-Analysis
EligMeta integrates automated trial discovery with eligibility-aware meta-analysis for precise, reproducible clinical evidence synthesis in precision medic...