saev - Sparse Auto-Encoders for Vision

This repo is a fork of saev - Sparse Auto-Encoders for Vision for use in the Imageomics Conference Tooling Workshop. Only minor modifications are made to help with facilitation. Overall functionality remains the same.

Training sparse autoencoders (SAEs) on vision transformers (ViTs), implemented in PyTorch.

Docs

• Docs
• Colab Notebook for SAE Inference
• User guide
• API reference

Research

This framework was developed for a series of projects leveraging SAEs with vision models.

• [Feb 2025] Interpretable and Testable Vision Features via Sparse Autoencoders
• [Nov 2025] Towards Open-Ended Visual Scientific Discovery with Sparse Autoencoders

Trained SAE checkpoints are available at:

• Huggingface Models

If you want to cite the software, please cite it as:

@software{stevens2025saev,
  author = {Stevens, Samuel},
  license = {CC-BY-4.0},
  month = apr,
  title = {{saev}},
  url = {https://github.com/OSU-NLP-Group/saev},
  year = {2025}
}