Search papers, labs, and topics across Lattice.
1
0
3
LLMs can't easily forget things across languages because erasing knowledge in one language often breaks the model in another, unless you carefully target the right layers.
