6 papers
Look But Don't Touch with Sparse Autoencoders for Unlearning in Diffusion Models
Enrico Cassano, Riccardo Renzulli, Rayyan Ahmed +2
Sparse autoencoders (SAEs) have recently been proposed as interpretable tools for concept-level manipulation, under the assumption that isolated features can serve as controllable…
Semantic Adapter Routing with Fine-Tuning Task Embeddings
Enrico Cassano, MichaÅ Brzozowski, Michał Brzozowski +3
Parameter-efficient fine-tuning (PEFT) has led to model ecosystems in which a single backbone is paired with many task-specialized adapters. Given such a library, routing aims to s…
Reading the Finetuning Prior: Verbatim Content Recovery via Contrastive Decoding Diffing
MichaÅ Brzozowski, Zuzanna Dubanowska, Enrico Cassano +1
Narrowly finetuned language models memorize implanted content verbatim, but auditing what a deployed model has been taught, without access to its weights or training data, remains…
SAEmnesia: Erasing Concepts in Diffusion Models with Supervised Sparse Autoencoders
Enrico Cassano, Riccardo Renzulli, Marco Nurisso +3
Concept unlearning in diffusion models is hampered by feature splitting, where concepts are distributed across many latent features, making their removal challenging and computatio…
MedSAE: Dissecting MedCLIP Representations with Sparse Autoencoders
Riccardo Renzulli, Colas Lepoutre, Enrico Cassano +1
Artificial intelligence in healthcare requires models that are accurate and interpretable. We advance mechanistic interpretability in medical vision by applying Medical Sparse Auto…
When Does Pruning Benefit Vision Representations?
Enrico Cassano, Riccardo Renzulli, Andrea Bragagnolo +1
Pruning is widely used to reduce the complexity of deep learning models, but its effects on interpretability and representation learning remain poorly understood. This paper invest…