README.md

metadata

license: cc-by-nc-4.0
library_name: saelens

Gemma Scope:

This is a landing page for Gemma Scope, a comprehensive, open suite of sparse autoencoders for Gemma 2 9B and 2B. Sparse Autoencoders are a "microscope" of sorts that can help us break down a model’s internal activations into the underlying concepts, just as biologists use microscopes to study the individual cells of plants and animals.

Key links:

• Learn more about Gemma Scope in our Google DeepMind blog post.
• Check out the interactive Gemma Scope demo made by Neuronpedia.
• Check out our Google Colab notebook tutorial for how to use Gemma Scope.
• Read the Gemma Scope technical report.
• Check out Mishax, a GDM internal tool that we used in this project to expose the internal activations inside Gemma 2 models.

Quick start:

You can get started with Gemma Scope by downloading the weights from any of our repositories:

• https://huggingface.co/google/gemma-scope-2b-pt-res
• https://huggingface.co/google/gemma-scope-2b-pt-mlp
• https://huggingface.co/google/gemma-scope-2b-pt-att
• https://huggingface.co/google/gemma-scope-2b-pt-transcoders
• https://huggingface.co/google/gemma-scope-9b-pt-res
• https://huggingface.co/google/gemma-scope-9b-pt-mlp
• https://huggingface.co/google/gemma-scope-9b-pt-att
• https://huggingface.co/google/gemma-scope-9b-it-res
• https://huggingface.co/google/gemma-scope-27b-pt-res

The full list of SAEs we trained at which sites and layers are linked from the following table, adapted from Figure 1 of our technical report:

Gemma 2 Model	SAE Width	Attention	MLP	Residual	Tokens
2.6B PT (26 layers)	2^14 ≈ 16.4K	All	All+	All	4B
	2^15			{12}	8B
	2^16	All	All	All	8B
	2^17			{12}	8B
	2^18			{12}	8B
	2^19			{12}	8B
	2^20 ≈ 1M			{5, 12, 19}	16B
9B PT (42 layers)	2^14	All	All	All	4B
	2^15			{20}	8B
	2^16			{20}	8B
	2^17	All	All	All	8B
	2^18			{20}	8B
	2^19			{20}	8B
	2^20			{9, 20, 31}	16B
27B PT (46 layers)	2^17			{10, 22, 34}	8B
9B IT (42 layers)	2^14			{9, 20, 31}	4B
	2^17			{9, 20, 31}	8B