List your company
ResearchFormerly WizWand

Machine Unlearning on ToFU NEK (ROUGE-L, Acc)

0.76ROUGE-LIHL + CE
0.09440.26720.440.6128Oct 20, 2025
Updated 6mo ago

Evaluation results

MethodLinks
IHL + CETraining Protocol=Adapters, Training Data Volume Parity=True2025.100.7677
IHL + CEtraining=adapters only2025.100.7373
AStraining=adapters only2025.100.7376
ASTraining Protocol=Adapters, Training Data Volume Parity=True2025.100.7173
IHL + KLTraining Protocol=Adapters, Training Data Volume Parity=True2025.100.6973
GA + CEtraining=adapters only2025.100.6468
IHL + KLtraining=adapters only2025.100.660
GA + CETraining Protocol=Adapters, Training Data Volume Parity=True2025.100.5957
ULDTraining Protocol=Adapters, Training Data Volume Parity=True2025.100.5761
NPO + KLtraining=adapters only2025.100.5555
ULDtraining=adapters only2025.100.5556
NPO + CEtraining=adapters only2025.100.435
NPO + CETraining Protocol=Adapters, Training Data Volume Parity=True2025.100.3836
IHLTraining Protocol=Adapters, Training Data Volume Parity=True2025.100.3845
NPO + KLTraining Protocol=Adapters, Training Data Volume Parity=True2025.100.3733
IHLtraining=adapters only2025.100.3233
GA + KLtraining=adapters only2025.100.2324
NPOtraining=adapters only2025.100.2326
NPOTraining Protocol=Adapters, Training Data Volume Parity=True2025.100.2120
GA + KLTraining Protocol=Adapters, Training Data Volume Parity=True2025.100.1919
GAtraining=adapters only2025.100.1723
GATraining Protocol=Adapters, Training Data Volume Parity=True2025.100.1218