CVPR 2025 Open Access Repository

Mechanistic Interpretability for Vision

Language-Guided Trajectory Traversal in Disentangled Stable Diffusion Latent Space for Factorized Medical Image Generation: Zahra TehraniNasab,

Amar Kumar,

Tal Arbel; [pdf] [supp] [arXiv]
[bibtex]
@InProceedings{TehraniNasab_2025_CVPR, author = {TehraniNasab, Zahra and Kumar, Amar and Arbel, Tal}, title = {Language-Guided Trajectory Traversal in Disentangled Stable Diffusion Latent Space for Factorized Medical Image Generation}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4846-4851} }
Embedding Shift Dissection on CLIP: Effects of Augmentations on VLM's Representation Learning: Ashim Dahal,

Saydul Akbar Murad,

Nick Rahimi; [pdf] [supp]
[bibtex]
@InProceedings{Dahal_2025_CVPR, author = {Dahal, Ashim and Murad, Saydul Akbar and Rahimi, Nick}, title = {Embedding Shift Dissection on CLIP: Effects of Augmentations on VLM's Representation Learning}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4814-4818} }
Leveraging Vision-Language Foundation Models to Reveal Hidden Image-Attribute Relationships in Medical Imaging: Amar Kumar,

Anita Kriz,

Barak Pertzov,

Tal Arbel; [pdf] [arXiv]
[bibtex]
@InProceedings{Kumar_2025_CVPR, author = {Kumar, Amar and Kriz, Anita and Pertzov, Barak and Arbel, Tal}, title = {Leveraging Vision-Language Foundation Models to Reveal Hidden Image-Attribute Relationships in Medical Imaging}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4840-4845} }
Disentangling Polysemantic Channels in Convolutional Neural Networks: Robin Hesse,

Jonas Fischer,

Simone Schaub-Meyer,

Stefan Roth; [pdf] [supp] [arXiv]
[bibtex]
@InProceedings{Hesse_2025_CVPR, author = {Hesse, Robin and Fischer, Jonas and Schaub-Meyer, Simone and Roth, Stefan}, title = {Disentangling Polysemantic Channels in Convolutional Neural Networks}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4799-4803} }
Uncovering Branch-specialization in InceptionV1 using k sparse autoencoders: Matthew Bozoukov; [pdf] [arXiv]
[bibtex]
@InProceedings{Bozoukov_2025_CVPR, author = {Bozoukov, Matthew}, title = {Uncovering Branch-specialization in InceptionV1 using k sparse autoencoders}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4809-4813} }
Wavelet-Based Mechanistic Interpretability of Vision Transformers via Frequency-Aware Ablations: Sophia J. Abraham,

Jonathan D. Hauenstein,

Walter J. Scheirer; [pdf]
[bibtex]
@InProceedings{Abraham_2025_CVPR, author = {Abraham, Sophia J. and Hauenstein, Jonathan D. and Scheirer, Walter J.}, title = {Wavelet-Based Mechanistic Interpretability of Vision Transformers via Frequency-Aware Ablations}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4830-4834} }
Visualizing and Controlling Cortical Responses Using Voxel-Weighted Activation Maximization: Matthew Shinkle,

Mark D. Lescroart; [pdf] [supp] [arXiv]
[bibtex]
@InProceedings{Shinkle_2025_CVPR, author = {Shinkle, Matthew and Lescroart, Mark D.}, title = {Visualizing and Controlling Cortical Responses Using Voxel-Weighted Activation Maximization}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4825-4829} }
Investigating Mechanisms for In-Context Vision Language Binding: Darshana Saravanan,

Makarand Tapaswi,

Vineet Gandhi; [pdf] [arXiv]
[bibtex]
@InProceedings{Saravanan_2025_CVPR, author = {Saravanan, Darshana and Tapaswi, Makarand and Gandhi, Vineet}, title = {Investigating Mechanisms for In-Context Vision Language Binding}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4852-4856} }
Naturally Computed Scale Invariance in the Residual Stream of ResNet18: André Longon; [pdf] [supp] [arXiv]
[bibtex]
@InProceedings{Longon_2025_CVPR, author = {Longon, Andr\'e}, title = {Naturally Computed Scale Invariance in the Residual Stream of ResNet18}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4804-4808} }
Analyzing Hierarchical Structure in Vision Models with Sparse Autoencoders: Matthew Lyle Olson,

Musashi Hinck,

Neale Ratzlaff,

Changbai Li,

Phillip Howard,

Vasudev Lal,

Shao-Yen Tseng; [pdf] [arXiv]
[bibtex]
@InProceedings{Olson_2025_CVPR, author = {Olson, Matthew Lyle and Hinck, Musashi and Ratzlaff, Neale and Li, Changbai and Howard, Phillip and Lal, Vasudev and Tseng, Shao-Yen}, title = {Analyzing Hierarchical Structure in Vision Models with Sparse Autoencoders}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4835-4839} }
Decoding Vision Transformers: the Diffusion Steering Lens: Ryota Takatsuki,

Sonia Joseph,

Ippei Fujisawa,

Ryota Kanai; [pdf] [supp] [arXiv]
[bibtex]
@InProceedings{Takatsuki_2025_CVPR, author = {Takatsuki, Ryota and Joseph, Sonia and Fujisawa, Ippei and Kanai, Ryota}, title = {Decoding Vision Transformers: the Diffusion Steering Lens}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {4819-4824} }