List your company

Harmful Content Detection on PHEME New Attacks: ExplainDrive (test)

82.91AccuracyLLM-SGA/ARHOCD
73.13475.67278.2180.748Dec 19, 2025

Evaluation results

MethodLinks
LLM-SGA/ARHOCDapproximation=Monte Carlo2025.1282.9181.6782.6182.0313.16
LLM-SGA/ARHOCDapproximation=Lower-bound2025.1282.8881.6482.588213.24
LLM-purimode=Enhanced2025.1282.4181.4180.6581.0214.24
MRATmode=Enhanced2025.1282.0581.2279.9180.4415.38
ATmode=Enhanced2025.1280.5779.6378.1978.7616.29
Det&Resmode=Enhanced2025.1280.0578.9377.8978.3216.27
EnsSelmode=Enhanced2025.1279.979.6876.4277.4318.44
OutRegmode=Enhanced2025.1279.8879.3676.7177.5917.52
ARTextmode=Enhanced2025.1279.1677.7477.8277.7715.11
ARDELmode=Enhanced2025.1279.0478.2975.9476.7316.29
FTMLmode=Enhanced2025.1278.1376.9875.475.9919.2
SAFERmode=Enhanced2025.1277.2775.7875.275.4517.54
MASKFilmode=Enhanced2025.1277.1575.7774.6175.0618.69
FATmode=Enhanced2025.1273.5973.3268.1768.9426.42
FIMmode=Enhanced2025.1273.5173.0568.2168.9825.36