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

Harmlessness evaluation on SafeEdit

99.26Success Rate

DINM

Updated 1mo ago

Evaluation Results

Method	Links
DINM 2026.02		99.26	99.56	99.56	99.11	99.85	1.5
PromptCD 2026.02		96.67	98.67	82.3	96.59	82.37	7.29
PromptCD 2026.02		95.26	92.96	84.81	95.78	83.85	5.95
DINM 2026.02		90	89.93	59.26	87.41	57.63	6.61
PromptCD 2026.02		76	88.89	76.89	77.78	76.96	6.44
FT-L 2026.02		42.52	76.22	28.37	42.89	28.85	5.94
Vanilla 2026.02		40.59	76.96	22.41	41.33	21.78	7.01
PromptCD 2026.02		38.67	86.67	36.44	39.78	37.04	7.14
DINM 2026.02		35.78	73.48	26.3	33.78	24.89	6.82
DINM 2026.02		26.74	84.22	34.59	24.52	33.78	7.78
Vanilla 2026.02		17.63	72.59	7.11	10.44	7.04	8.01
FT-L 2026.02		12.3	74	24.59	11.26	24.89	7.69
Vanilla 2026.02		12.15	54.81	18.07	12.07	18.74	7.46
FT-L 2026.02		11.7	54.96	30.59	11.78	30.52	7.16
FT-L 2026.02		10.67	36.67	19	10.22	18.44	6.65
Vanilla 2026.02		9.85	34.44	9.04	9.78	8.44	7.66