File size: 1,119 Bytes
fea284f 8f91220 44dcfe5 fea284f a623e7d 0403e97 8f91220 fea284f 8f91220 fea284f 8f91220 fea284f 8f91220 fea284f 8f91220 fea284f 8f91220 fea284f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 |
---
license: cc-by-4.0
library_name: saelens
---
! WARNING: we are in the process of uploading the 9B SAEs, and expect the initial set to land in 60 mins.
In the mean time the 2B SAEs, e.g. https://huggingface.co/google/gemma-scope-2b-pt SAEs should be available.
# 1. Gemma Scope
Gemma Scope is 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.
See our [landing page](https://huggingface.co/google/gemma-scope) for details on the whole suite. This is a specific set of SAEs:
# 2. What Is `gemma-scope-9b-pt-mlp`?
- `gemma-scope-`: See 1.
- `9b-pt-`: These SAEs were trained on Gemma v2 9B base model.
- `mlp`: These SAEs were trained on the model's MLP sublayer outputs.
## 3. Point of Contact
Point of contact: Arthur Conmy
Contact by email:
```python
''.join(list('moc.elgoog@ymnoc')[::-1])
```
HuggingFace account:
https://huggingface.co/ArthurConmyGDM
|