Updated 10 months ago

arrakis-mi • Science 44%

Arrakis is a library to conduct, track and visualize mechanistic interpretability experiments.

Updated 10 months ago

mlsae • Science 41%

Multi-Layer Sparse Autoencoders (ICLR 2025)

Updated 10 months ago

understanding-clip-ood • Science 36%

Official code for the paper: "When and How Does CLIP Enable Domain and Compositional Generalization?" (ICML 2025 Spotlight)