Search papers, labs, and topics across Lattice.
1
0
Forgetting in language models can be optimized at the token level, leading to state-of-the-art performance in unlearning without the need for external supervision.
