Pytorch implementation of 'Nonlinear Concept Erasure: A Density Matching Approach' (Saillenfest & Lemberger, 2025), Proceedings of ECAI 2025 - 28th European conference on AI
MUtE is a framework for erasing sensitive concepts from continuous representations while inducing a deterministic counterfactual mapping, which allows for both bias mitigation and the generation of counterfactuals
Domain-adapted OCR for multilingual medical documents: LoRA/PEFT adaptation of compact VLMs, seed-level reproducibility and an evaluation gate, the OCR error cascade under shortcut control, and redaction mechanisms (masking, adversarial gate, LEACE) under probe and inversion attacks. Paper code + Zenodo data (10.5281/zenodo.22078532).
Bias & Safety Auditing Pipeline for Text-to-Image models (Stable Diffusion). Detects demographic bias via CLIP, filters NSFW content, erases harmful concepts via attention fine-tuning, and auto-generates a PDF audit report. CAP6412 - UCF CRCV, Spring 2025.
To associate your repository with the concept-erasure topic, visit your repo's landing page and select "manage topics."