MUSE-AI-Lab / Concept-Brittleness
git clone https://github.com/MUSE-AI-Lab/Concept-Brittleness.git
Code for diagnosing and repairing object-dependent concept failures in text-to-image diffusion models using per-step sparse autoencoders.
Why you might care
Researchers studying why text-to-image models fail on specific objects for a fixed style concept can use it to locate missing or delayed concept features. It adds an inference-time correction toward a class-level prototype without retraining the diffusion model.
+0
stars, 24h · #68,411
+262
stars, 7 days · #1,129
+262
stars, 30 days · #3,946
262
★ total · +262% this week
0
forks · 6 open issues
Star history (as recorded here)
LanguagePython
License—
Created2026-10-28 (5d ago)
Last push5d ago
Watchers0
OwnerOrganization
Links
README
Loading README…
Blurb generated by claude-haiku-5-5 3d ago. Star deltas marked ~ are extrapolated from partial history.