ai-safety-foundation / sparse_autoencoder

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

Don't rely on pwd for setting up .checkpoints #205

Closed JasonGross closed 8 months ago

JasonGross commented 8 months ago

Fix #204