Share your thoughts, 1 month free Claude Pro on usSee more

Classification on MLLMU-Bench Forget Set

63.33Accuracy

Base

Updated 1d ago

Evaluation Results

Method	Links
Base 2026.05		63.33
GA 2026.05		61.67
NPO 2026.05		61.67
GA_Diff 2026.05		60.83
KL_Min 2026.05		60
PO 2026.05		60
Noise-T 2026.05		54.17
VGID 2026.05		50
ICU-T 2026.05		49.17
Base 2026.05		49.17
ICU-T 2026.05		39.17
KL_Min 2026.05		38.33
PO 2026.05		38.33
GA 2026.05		37.5
NPO 2026.05		37.5
RMU 2026.05		36.67
GA_Diff 2026.05		36.67
Noise-T 2026.05		36.67
RMU 2026.05		35.83
VGID 2026.05		35
PO 2026.05		34.17
RMU 2026.05		32.5
GA 2026.05		31.67
Base 2026.05		28.33
NPO 2026.05		28.33
Noise-T 2026.05		27.5
ICU-T 2026.05		25.83
KL_Min 2026.05		25.33
VGID 2026.05		24.17
GA_Diff 2026.05		18.83
Vanilla 2025.02		0.5187
Vanilla 2025.02		0.517
Prompting 2025.02		0.4973
Vanilla 2025.02		0.4915
Prompting 2025.02		0.4841
KL Minimization 2025.02		0.476
NPO 2025.02		0.474
KL Minimization 2025.02		0.468
Prompting 2025.02		0.468
NPO 2025.02		0.4561
NPO 2025.02		0.4552
KL Minimization 2025.02		0.448
GA 2025.02		0.444
GA 2025.02		0.4385
Gradient Difference 2025.02		0.436
Gradient Difference 2025.02		0.4347
MANU 2025.02		0.4205
Gradient Difference 2025.02		0.416
MANU 2025.02		0.4125
GA 2025.02		0.4093
MANU 2025.02		0.3836