Ashkan Khakzar

Learnable Sparsity for Vision Generative Models

Towards Understanding Multimodal Fine-Tuning: A Case Study into Spatial Features

Measuring Sparse Autoencoder Feature Space Similarities Across Large Language Models

Mixture of Experts Made Intrinsically Interpretable

Too Late to Recall: Explaining the Two-Hop Problem in Multimodal Knowledge Retrieval

Minimalist Concept Erasure in Generative Models

Learning Visual Prompts for Guiding the Attention of Vision Transformers