Multimodal Machine Unlearning on MLLMU-Bench benchmark leaderboard · SOTA2 Research

Multimodal Machine Unlearning on MLLMU-Bench

29.6Forget VQA AccuracyCVF

Evaluation Results

Method	Links
CVF2026.05		29.6	38.4	45	37.5	47	54.9
MMU2026.05		31.2	34.2	44.2	35.1	46.7	54.9
GA_Diff2026.05		40	33.6	44.3	31.5	46.6	53.6
MANU2026.05		41.6	33.6	43.2	32.1	46.3	54.8
KL_Min2026.05		42.4	33.6	44.9	32	47.4	54.6
CVF2026.05		42.8	55	56	58.6	75.4	77.5
GA2026.05		43.2	32.5	45	32.2	47	54.7
NPO2026.05		43.2	33.6	44.9	32.2	47	54.7
MANU2026.05		43.2	47.2	43.2	49.4	73.7	76.2
MMU2026.05		44	54.4	47.4	55.7	75.2	77.3
KL_Min2026.05		45.6	45.3	35.9	55.6	74.8	77.1
Vanilla2026.05		45.8	38.4	45.2	37.5	47.4	54.9
NPO2026.05		49.6	50.4	49.5	53.3	75.2	78.3
GA2026.05		50.4	46.7	51.5	57.6	74.4	77.8
GA_Diff2026.05		54.4	52.8	38.8	54.4	74.5	77
Vanilla2026.05		55.2	55	56	58.6	77.3	77.5