How to choose a vision model for your specific needs? How do ConvNet / ViT, supervised / CLIP models compare with each other on metrics beyond ImageNet? Our work comprehensively compares common vision models on "non-standard" metrics. (1/n)

Can we understand & edit unanticipated mechanisms in LMs? We introduce sparse feature circuits, & use them to explain LM behaviors, discover & fix LM bugs, & build an automated interpretability pipeline! Preprint w/ Can Rager, Eric J. Michaud, Yonatan Belinkov, David Bau, Aaron Mueller

📢 Recruiting participants for a paid study about LLMs I'm looking for US-based adults for a 1 hour study on Zoom (payment: $15 Amazon gift card), especially those with technical knowledge of LLMs Sign up at forms.gle/6aGfJNrfhpuQzm… Please RT and help out a grad student 😁

⏱️Few days left to submit your work to our #ECCV2024 Workshop on Uncertainty Quantification for Computer Vision European Conference on Computer Vision #ECCV2024 We welcome contributions on different facets of reliability of perception models. ⏲️Deadline: Wed 17th July 🎁 Bonus: we have an excellent speaker panel

Did you know that networks trained with different learning rates extract different features (and a different number of them!) from the data? Come by our poster at HiLD Workshop #ICML2024 tomorrow to discuss it with Ildus Sadrtdinov! Paper: openreview.net/forum?id=IID2D… 1/6

Representation learning is often done by considering samples to be either identical (same class, positive pairs) or not–with no middle ground. We propose 𝕏-CLR to learn from soft inter-sample relationships, and get better accuracy & improved robustness. arxiv.org/abs/2407.18134