Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Difference

Devs

Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Difference | Read Paper on Bytez