Downloads · 30 days
0
SAE-Phi-3-mini-4k-instruct-5e0-3-L1
analist/SAE-Phi-3-mini-4k-instruct-5e0-3-L1
SAE-Phi-3-mini-4k-instruct-5e0-3-L1 is a machine learning model from analist. 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.
These are trained Sparse Autoencoders (SAEs) for the microsoft/Phi-3-mini-4k-instruct model, compatible with the saelens library.
Access
Public
Updated Jun 9, 2026
Repo size
1.8 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors1.8 GB · 100%
At a glance
- Library
- saelens
- Access
- Public
- Created
- Jun 9, 2026
- Updated
- Jun 9, 2026
- SHA
- 0acb5245
Model card
From the Hugging Face model README
Sparse Autoencoders for Phi-3-mini-4k-instruct
These are trained Sparse Autoencoders (SAEs) for the microsoft/Phi-3-mini-4k-instruct model, compatible with the sae_lens library.
Repository Structure
layer_{L}/- Contains the final SAELens-compatible SAFETENSORS for each target layer.
Model Details
- Model:
microsoft/Phi-3-mini-4k-instruct - Layers: [21, 22, 23]
- Architecture: Standard (ReLU) with pre-encoder centring (
b_decapplied to input). - Expansion Factor: 16x (49152 features)
- Tokens Trained: 645,832,704
Datasets
- NuminaMath-CoT: 40%
- HH-RLHF: 20%
- FineWeb: 20%
- JBB/HarmBench/AdvBench: 20%
Usage with SAELens
from sae_lens import SAE
# Load the final SAE for a specific layer, e.g., layer 21
sae, cfg_dict, sparsity = SAE.from_pretrained(
release="analist/SAE-Phi-3-mini-4k-instruct-5e0-3-L1",
sae_id="layer_21",
device="cuda"
)
- Library
- saelens
- Created
- Jun 9, 2026
- Updated
- Jun 9, 2026