Downloads · 30 days
51
12% of all-time downloads
nanointerpret-qwen3
Belluxx/nanointerpret-qwen3
nanointerpret-qwen3 is a machine learning model from Belluxx. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as mit.
A 32,768-feature Top-K SAE trained on the layer 14 residual stream of Qwen3 1.7B Base using 500M tokens from FineWeb-Edu. Includes checkpoint, 100M tokens of recorded activations, and feature interpretations.
All-time downloads
436
Public
Repo size
21 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.npy10 GB · 95%
At a glance
- License
- mit
- Access
- Public
- Created
- Aug 17, 2026
- Updated
- Aug 27, 2026
- SHA
- 77260344
Base models
Datasets
Model card
From the Hugging Face model README
Qwen3 1.7B sparse autoencoder
A 32,768-feature Top-K SAE trained on the layer 14 residual stream of Qwen3 1.7B Base using 500M tokens from FineWeb-Edu. Includes checkpoint, 100M tokens of recorded activations, and feature interpretations.
Explore
Browse the features at nanointerpret.pages.dev, or run the visualizer locally:
git clone https://github.com/Belluxx/nanointerpret
cd nanointerpret
pip install -r requirements.txt
hf download Belluxx/nanointerpret-qwen3 --local-dir artifacts/nanointerpret-qwen3
python visualize.py --activations artifacts/nanointerpret-qwen3/activations
Training code and details: Belluxx/nanointerpret.
Metrics
| Width / K | 32,768 / 16 |
| Explained variance | 71.88% |
| Validation MSE | 0.1558 |
| Downstream KL | 0.2048 |
| Dead features | 1.74% |
- License
- mit
- Created
- Aug 17, 2026
- Updated
- Aug 27, 2026