OPTML-Group/Unlearn-Simple
[NeurIPS25] Official repo for "Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning"
This project helps machine learning engineers and researchers modify large language models (LLMs) to remove specific, unwanted information or capabilities without the extensive cost of retraining the entire model. It takes an existing LLM and a dataset of 'forget' content (like copyrighted material or harmful patterns) as input, producing an 'unlearned' LLM that no longer exhibits the unwanted behaviors while retaining its general usefulness. This is for AI developers, MLOps engineers, and data scientists working with LLMs.
No commits in the last 6 months.
Use this if you need to quickly and efficiently remove specific harmful, biased, or proprietary information from a deployed large language model without rebuilding it from scratch.
Not ideal if you are looking to fine-tune an LLM for new capabilities or entirely retrain a model from the ground up for a different purpose.
Stars
43
Forks
12
Language
Python
License
MIT
Category
Last pushed
Oct 03, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/OPTML-Group/Unlearn-Simple"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
agentscope-ai/Trinity-RFT
Trinity-RFT is a general-purpose, flexible and scalable framework designed for reinforcement...
OpenRLHF/OpenRLHF
An Easy-to-use, Scalable and High-performance Agentic RL Framework based on Ray (PPO & DAPO &...
zjunlp/EasyEdit
[ACL 2024] An Easy-to-use Knowledge Editing Framework for LLMs.
huggingface/alignment-handbook
Robust recipes to align language models with human and AI preferences
hyunwoongko/nanoRLHF
nanoRLHF: from-scratch journey into how LLMs and RLHF really work.