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

Multimodal Machine Unlearning Evaluation on MLLMU-Bench (test)

47.86Classification Accuracy

Vanilla

33.66437.349541.03544.7205Feb 21, 2025May 15, 2025Aug 7, 2025Oct 29, 2025Jan 21, 2026Apr 14, 2026Jul 7, 2026
Updated 16d ago

Evaluation Results

MethodLinks
2025.11
47.860.5394.8923.01
2025.11
47.530.5024.0825.33
2025.11
47.410.515.225.43
2025.11
47.290.4794.2124.11
2025.11
46.810.4833.6724.56
2025.11
46.420.4084.2521.66
2025.11
45.20.3964.5420.04
2025.11
44.870.4154.1821.99
2025.11
44.440.3473.9120
2025.11
43.950.3583.8419.35
2026.07
43.520.5165.225.73
2025.11
43.410.3833.8316.19
2025.11
43.20.4393.7821.09
2026.07
43.050.4513.820.8
2026.07
42.90.4614.7218.2
2025.11
42.750.423.2920.5
2026.07
42.750.423.2920.5
2025.11
42.670.3313.7218.81
2025.11
42.180.4013.6118.11
2026.07
41.70.4753.5517.46
2025.11
40.870.4323.3516.92
2025.11
40.60.4213.1915.77
2026.07
40.20.3873.8314.51
2025.11
40.150.4453.5217.88
2025.02
39.750.3553.5121.88
2025.11
39.640.3713.717.67
2025.11
39.330.4394.0117.88
2025.11
39.080.4143.0714.5
2026.07
39.080.4143.0714.5
2025.11
38.40.3843.4716.47
2025.11
38.320.4213.0817.11
2025.02
35.550.3613.9120.97
2025.02
35.320.3333.116.66
2025.11
34.210.4073.0119.78