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

CADD

Benchmarks

Task NameDataset NameSOTA ResultTrend
Harmful content detectionCADD
Accuracy90.19
8
Harmful content detectionCADD DeepSeek generations
Accuracy60.13
4
Harmful content detectionCADD Llama-3.1 generations
Accuracy69.18
4
Harmful content detectionCADD GPT-4o generations
Accuracy64.84
4
Toxic-neutral pair quality evaluationTranslated CADD
Overall Score2.963
1
Showing 5 of 5 rows