jinzhuoran/RWKU
RWKU: Benchmarking Real-World Knowledge Unlearning for Large Language Models. NeurIPS 2024
This project helps AI researchers and machine learning engineers evaluate how effectively large language models (LLMs) can forget specific real-world information, such as facts about famous people. You provide an LLM and the knowledge you want it to forget (e.g., "Stephen King"), and the project measures if the model successfully unlearned that information without affecting its other abilities. The end-user persona is an AI researcher, LLM developer, or machine learning engineer focused on model safety and ethical AI.
No commits in the last 6 months.
Use this if you need a standardized benchmark to rigorously test and compare different knowledge unlearning methods for large language models.
Not ideal if you are looking for a tool to implement a knowledge unlearning technique rather than evaluate one.
Stars
91
Forks
9
Language
Python
License
—
Category
Last pushed
Sep 30, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/jinzhuoran/RWKU"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
steering-vectors/steering-vectors
Steering vectors for transformer language models in Pytorch / Huggingface
jianghoucheng/AlphaEdit
AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models, ICLR 2025 (Outstanding Paper)
kmeng01/memit
Mass-editing thousands of facts into a transformer memory (ICLR 2023)
boyiwei/alignment-attribution-code
[ICML 2024] Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications
jianghoucheng/AnyEdit
AnyEdit: Edit Any Knowledge Encoded in Language Models, ICML 2025