Downloads · 30 days
0
tetrahedron-saes-per-head-attn
timaeus/tetrahedron-saes-per-head-attn
tetrahedron-saes-per-head-attn is a machine learning model from timaeus. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for saelens.
This repository contains the following SAEs: - blocks.0:0.attn.hookz - blocks.0:1.attn.hookz - blocks.0:2.attn.hookz - blocks.0:3.attn.hookz - blocks.0:4.attn.hookz - blocks.0:5.attn.hookz - blocks.0:6.attn.hookz - bl…
Access
Public
Updated Sep 19, 2024
Repo size
2.2 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors2.2 MB · 98%
At a glance
- Library
- saelens
- Access
- Public
- Created
- Sep 19, 2024
- Updated
- Sep 19, 2024
- SHA
- 692d6109
Model card
From the Hugging Face model README
SAEs for use with the SAELens library
This repository contains the following SAEs:
- blocks.0:0.attn.hook_z
- blocks.0:1.attn.hook_z
- blocks.0:2.attn.hook_z
- blocks.0:3.attn.hook_z
- blocks.0:4.attn.hook_z
- blocks.0:5.attn.hook_z
- blocks.0:6.attn.hook_z
- blocks.0:7.attn.hook_z
- blocks.1:0.attn.hook_z
- blocks.1:1.attn.hook_z
- blocks.1:2.attn.hook_z
- blocks.1:3.attn.hook_z
- blocks.1:4.attn.hook_z
- blocks.1:5.attn.hook_z
- blocks.1:6.attn.hook_z
- blocks.1:7.attn.hook_z
Load these SAEs using SAELens as below:
from sae_lens import SAE
sae, cfg_dict, sparsity = SAE.from_pretrained("timaeus/tetrahedron-saes-per-head-attn", "<sae_id>")
- Library
- saelens
- Created
- Sep 19, 2024
- Updated
- Sep 19, 2024