Back
Scale Alone Does not Improve Mechanistic Interpretability in Vision Models
@inproceedings{zimmermann2023scale, title = {Scale Alone Does not Improve Mechanistic Interpretability in Vision Models}, booktitle = {Advances in Neural Information Processing Systems 36 (NeurIPS 2023)}, pages = {57876 - 57907}, address = {Curran Associates Inc.}, month = oct, year = {2023}, slug = {zimmermann2023scale}, author = {Zimmermann, Roland S and Klein, Thomas and Brendel, Wieland}, url = {https://papers.nips.cc/paper_files/paper/2023/file/b4aadf04d6fde46346db455402860708-Paper-Conference.pdf}, month_numeric = {10} }
More information