Knowledge Retention on Comprehensive Knowledge Retention Suite (MME, MMBench, MMMU, POPE) benchmark leaderboard · SOTA2 Research

Knowledge Retention on Comprehensive Knowledge Retention Suite (MME, MMBench, MMMU, POPE)

65.61COMLLaVA-v1.5 (7B)

Evaluation Results

Method	Links
LLaVA-v1.5 (7B)2025.10		65.61	45.59	49.22	66.33	26.37	19.33	54.32	46.74	26.93	—
O-LoRA2025.10		61.47	40.91	48.07	34.85	17.28	19.87	51.12	41.25	25.26	36.7
MoELoRA2025.10		60.79	38.79	48.27	35.03	17.85	19.79	49.99	40.51	24.92	37.22
Replay2025.10		59.72	37.98	48.64	62.33	19.31	19.17	51.67	43	28.83	66.04
LwF2025.10		53.14	28.77	43.41	36.19	13.68	18.22	44.18	35.44	26.36	66.26
KORE2025.10		52.48	39.96	48.96	60.02	23.18	18.09	51.5	42.1	39.14	84.93
CIA2025.10		52.47	33.8	45.09	34.07	10.4	12.5	44.52	35.99	26.69	66.86
KORE2025.10		52.41	40.98	48.68	38.54	16.58	18.59	51.75	40	37.98	82.81
EWC2025.10		49.42	32.88	45.46	29.79	13.36	18	43.5	35.14	26.3	66.26
LoRA2025.10		48.96	27.01	43.79	29.66	13.7	18.02	41.38	33.47	25.12	63.83
Full-FT2025.10		43.55	21.55	45.67	25.25	13.03	18.32	16.09	31.66	24.13	61.98
SEFE2025.10		36.77	35.94	48.55	42.9	10.55	18.59	31.55	33.02	26.6	67.56