Reproduce “Machine Unlearning for Large Language Models: Foundations, Advances, and Agentic Extensions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiaoyu Xu, Minxin Du, Li Bai, Junxu Liu, Yaxin Xiao, Kun Fang, Liu Yang, Huadi Zheng and 3 more
The authors present a narrative review of machine unlearning in connected assistants, linking removal requests to scope, information location, intervention, and evidence-backed claims. They use seven lifecycle stages and six evidence dimensions. In an audit of 50 synthetic-canary histories, raw-only deletion left residue scores of 0.20/0.22, while full purge or tombstone redaction reached zero in tested settings, not universal removal.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.