TheMachine Press

A daily newspaper for the age of artificial intelligence.

Morning editionPermanent story

research

Distance Failed to Predict What the Model Would Relearn

FRAG scores whether an unlearning update targets forget-critical weights while sparing retain-critical ones.

Published Updated Story ID: mp-2026-08-27-009
Read the complete editionStory JSON

Summary

FRAG scores whether an unlearning update targets forget-critical weights while sparing retain-critical ones.

The authors argue that global weight displacement confuses selective unlearning with random or destructive change. Their training-free Forget-Retain Alignment Gap better separated selective from dense updates, and a pruning method built on the same principle improved relearning robustness in the reported experiments.

Why it matters

FRAG scores whether an unlearning update targets forget-critical weights while sparing retain-critical ones.

Limits and context

No additional limitation was separately recorded.

Key claims

  1. FRAG scores whether an unlearning update targets forget-critical weights while sparing retain-critical ones.

    Evidence: source-2026-08-27-009

Sources

  1. arXiv preprint 2608.25429arXiv · primary research

Corrections

No corrections have been recorded for this story.