Cataloging 527 Ways to Make LLMs Forget

What it does

This repository is a curated bibliography—an “awesome-list”—that tracks research on machine unlearning in large language models. It catalogs 527 papers, 17 surveys, 3 frameworks, and 2 blog posts, organized chronologically from 2021 through 2026. It serves as a community index for research on removing specific knowledge from already-trained models.

The interesting bit

Machine unlearning is the technical answer to the “right to be forgotten” for models that memorize their training corpora. The list tracks how the field has moved beyond privacy compliance into practical problems like copyright removal, hallucination suppression, and detoxification—essentially editing model weights after the fact.

Key highlights

• 527 papers spanning 2021–2026, plus dedicated sections for surveys, frameworks, and blog posts
• Coverage includes privacy, copyright, harmful content, hallucinations, and vision-language unlearning
• Entries note arXiv links, publication venues, and associated code repositories where available
• The maintainer explicitly solicits updates, corrections, and missing papers via issues and pull requests

Verdict

Worth bookmarking if you research AI safety, model editing, or data privacy in foundation models. Skip it if you are hunting for a unified unlearning framework or ready-to-run code—this is a reading list, not a library.