ai-safety-foundation / sparse_autoencoder

Sparse Autoencoder for Mechanistic Interpretability
https://ai-safety-foundation.github.io/sparse_autoencoder/
MIT License
171 stars 39 forks source link

Update the metrics to have a components dimension #162

Closed alan-cooney closed 8 months ago