Sign inSign in

Rongzhe Wei treats knowledge in large language models as correlated structure, not isolated atomic facts.

More

Rongzhe Wei treats knowledge in large language models as correlated structure, not isolated atomic facts. At research level, Wei presents a Correlated Knowledge Attack agent that splits harmful goals into harmless subqueries, plus a graph-based deep unlearning test.

Playlists

1

About

Rongzhe Wei treats knowledge in large language models as correlated structure, not isolated atomic facts. At research level, Wei presents a Correlated Knowledge Attack agent that splits harmful goals into harmless subqueries, plus a graph-based deep unlearning test.

Connections

3

Their places, courses, books and lectures.