YaNgZhAnG-V5/minimalist_concept_erasure
[ICML25] Minimalist Concept Erasure in Generative Models
This tool helps content creators and platforms ensure that images generated by AI models are free from unwanted elements like nudity, copyrighted characters, or specific artistic styles. You provide an existing text-to-image AI model and specify the concepts to remove, and it produces a modified model that avoids generating those concepts while keeping its general image creation abilities. It's for anyone who uses generative AI to produce visual content and needs to control its output for brand safety, legal compliance, or artistic consistency.
Use this if you need to clean up the output of your text-to-image AI models to prevent the generation of specific, undesirable content.
Not ideal if you are looking to create entirely new visual concepts or styles, as this tool is focused on removing existing ones.
Stars
16
Forks
—
Language
Python
License
—
Category
Last pushed
Dec 02, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/generative-ai/YaNgZhAnG-V5/minimalist_concept_erasure"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
soran-ghaderi/torchebm
🍓 Build and train energy-based and diffusion models in PyTorch ⚡.
spqb/adabmDCApy
Pytorch implementation of adabmDCA
opendilab/GenerativeRL
Python library for solving reinforcement learning (RL) problems using generative models (e.g....
AstraZeneca/DiffAbXL
The official implementation of DiffAbXL benchmarked in the paper "Exploring Log-Likelihood...
G-U-N/UniRL
a unified reinforcement learning toolbox for joint RL on language models and diffusion models