castorini/daam

Diffusion attentive attribution maps for interpreting Stable Diffusion.

51
/ 100
Established

Extracts cross-attention weights from the diffusion process to generate per-word attribution heatmaps, enabling fine-grained visualization of which image regions each token influences. Supports both Stable Diffusion and SDXL via the Hugging Face Diffusers library, with a Python API for programmatic use and CLI tools for quick generation and serialization of attribution experiments.

788 stars and 348 monthly downloads. No commits in the last 6 months. Available on PyPI.

Stale 6m No Dependents
Maintenance 0 / 25
Adoption 16 / 25
Maturity 18 / 25
Community 17 / 25

How are scores calculated?

Stars

788

Forks

69

Language

Jupyter Notebook

License

MIT

Last pushed

Apr 05, 2024

Monthly downloads

348

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/castorini/daam"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.