/ULD

Implementation of paper 'Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Difference' [NeurIPS'24]

Primary LanguagePythonMIT LicenseMIT

Stargazers