Search papers, labs, and topics across Lattice.
AIRI, Skoltech
1
0
3
Unlearning is much easier on supervised fine-tuned models than on pretrained ones, with direct unlearning on pretrained models often leading to catastrophic forgetting.
