WritingPrompts
Benchmarks
TaskDatasetSOTA resultTrendResults
WritingPrompts small Fast-DetectGPT benchmark (test)
99.9AUROC
54
WritingPrompts-small Fast-DetectGPT benchmark
99.9AUROC
54
WritingPrompts
32MAUVE
33
WritingPrompts
1AUROC
30
WritingPrompts
8.71LLM Diversity
23
WritingPrompts
100TPR (FPR=0.01)
18
WriPro (WritingPrompts)
1,813Topic Judge Elo Score
16
WritingPrompts
98.44Brier Accuracy (BA)
16
WritingPrompts
34.6ROUGE-1
15
WritingPrompts GPT-J-6B
97.6AUROC
15
WritingPrompts English (test)
98.8TPR@FPR5%
15
WritingPrompts (test)
88Diversity (div)
14
WritingPrompts
3.7Diversity
10
WritingPrompts
22.11F1 Score
10
WritingPrompts
1.76PPL
10
WritingPrompts (WP) (test)
0.224BLEU-1
10
WritingPrompts
2.78Rating
8
WritingPrompts
87.8BERTScore
8
WritingPrompts
20.64Character Development
8
WritingPrompts super-long sequences (> 17k tokens) OOD
195.89MAE
8
WritingPrompts
10.9Bad Output Rate
5
WritingPrompts (test)
8.14PPL (Generation)
5
WritingPrompts
94.23TPR
5
WritingPrompts Fast-DetectGPT
98.8AUROC
5
WritingPrompts
58.6BLEU
4