OpenMOSS / Language-Model-SAEs

For OpenMOSS Mechanistic Interpretability Team's Sparse Autoencoder (SAE) research.
21 stars 3 forks source link

Add Configuration Entrypoint #21

Closed dest1n1s closed 2 weeks ago

dest1n1s commented 2 weeks ago

This PR changed configs to compositional style. It includes: