WritingPrompts
Benchmarks
TaskDatasetSOTA resultTrendResults
WritingPrompts Llama3-8B
100TPR @ FPR=1%
3
WritingPrompts GPT-4.1-mini
31.33TPR @ FPR=1%
3
WritingPrompts cross-source corruption DetectRL (test)
84.15AUROC
3
WritingPrompts (WP) (test)
73.88Fascination
2
WritingPrompts (test)
85Same Count
2