PuSH - Publication Server of Helmholtz Zentrum München

Lim, H. ; Choi, J.* ; Choo, J.* ; Schneider, S.

Sparse autoencoders reveal selective remapping of visual concepts during adaptation.

In: (13th International Conference on Learning Representations Iclr 2025, 24 - 28 April 2025, Singapur). 2025. 46012-46037 (13th International Conference on Learning Representations Iclr 2025)
Postprint
Adapting foundation models for specific purposes has become a standard approach to build machine learning systems for downstream applications. Yet, it is an open question which mechanisms take place during adaptation. Here we develop a new Sparse Autoencoder (SAE) for the CLIP vision transformer, named PatchSAE, to extract interpretable concepts at granular levels (e.g., shape, color, or semantics of an object) and their patch-wise spatial attributions. We explore how these concepts influence the model output in downstream image classification tasks and investigate how recent state-of-the-art prompt-based adaptation techniques change the association of model inputs to these concepts. While activations of concepts slightly change between adapted and non-adapted models, we find that the majority of gains on common adaptation tasks can be explained with the existing concepts already present in the non-adapted foundation model. This work provides a concrete framework to train and use SAEs for Vision Transformers and provides insights into explaining adaptation mechanisms.
Additional Metrics?
[➜Log in]
Edit extra informations Login
Publication type Article: Conference contribution
Corresponding Author
ISSN (print) / ISBN [9798331320850]
Conference Title 13th International Conference on Learning Representations Iclr 2025
Conference Date 24 - 28 April 2025
Conference Location Singapur
Quellenangaben Volume: , Issue: , Pages: 46012-46037 Article Number: , Supplement: ,
Non-patent literature Publications
Institute(s) Institute of Computational Biology (ICB)