Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2

Meta info.

• Authors: Google Deepmind Language Model Interpretability team
• Paper: https://storage.googleapis.com/gemma-scope/gemma-scope-report.pdf
• Affiliation: Google DeepMind
• References: DeepMind Blog, gemma-scope HF

TL; DR

LM (Gemma 2) interpretability를 위한 Gemma Scope suite 공개에 따른 technical Report