Model Editing on CounterFact (Efficacy, Generalization, Specificity, Edited Retention) benchmark leaderboard · SOTA2 Research

Model Editing on CounterFact (Efficacy, Generalization, Specificity, Edited Retention)

73.06EfficacyAlphaEdit

Evaluation Results

Method	Links
AlphaEdit2026.04		73.06	66.43	48.54	47.4
HiEditfull2026.04		66.53	55.65	45.7	58.2
HiEditrand2026.04		66.4	55.48	45.51	58
RLEdit2026.04		66.35	55.26	44.79	57.2
FT2026.04		65.24	57.99	44.03	62.8
HiEditfull2026.04		64.06	52.8	43.59	54
FT2026.04		60.13	39.49	52.26	42.7
HiEditrand2026.04		59.81	50	47.37	50.3
ROME2026.04		59.26	56.45	48.52	48.3
RLEdit2026.04		55.51	49.92	46.28	45.8
AlphaEdit2026.04		38.17	38.83	66.51	34
PRUNE2026.04		32.41	29.78	65.1	29.8
HiEdit_full2026.04		31.66	22.41	10.8	23.05
MEMIT2026.04		31.51	28.78	65.58	30.1
HiEdit_rand2026.04		31.45	22.35	10.96	22.75
RECT2026.04		30.72	28.34	66.08	30
DAFNet2026.04		29.56	31.55	68.38	33.6
RLEdit2026.04		27.93	20.58	10.83	21
MEND2026.04		27.28	28.54	70.2	27.7
MEMIT2026.04		18.61	16.94	16.93	16.2
AlphaEdit2026.04		11.64	4.87	1.49	1.6
MALMEN2026.04		10.46	13.12	86.94	11.4
DAFNet2026.04		10.23	12.74	86.6	13.5
Pre-edited2026.04		8.23	10.8	88.72	10.4
MEND2026.04		7.69	13.4	7.45	10.5
Pre-edited2026.04		7.2	9.1	89.77	8.7
ROME2026.04		1.78	1.8	1.17	0.9
Pre-edited2026.04		1.01	1.01	21.74	1.35
FT2026.04		0.29	0.06	0	0.03
PRUNE2026.04		0	0	0	0
RECT2026.04		0	0	0	0
MALMEN2026.04		0	0	0	0
ROME2026.04		0	0	0	0