WIKIBIO
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
WikiBio
2.3AUC π-Soft-NS
67
WikiBio (test)
45.14BLEU
17
WikiBio NonFact Sentence-level
85.3AUC-PR
16
WikiBio
100Edit Success Rate
15
WikiBio GPT-4o
85Accuracy
12
WikiBIO
9.68BLEU
11
WikiBio Continual
35.5Soft Token Accuracy
10
WikiBio
35.5ROUGE-L
10
WikiBio
68.6AUROC
10
WIKIBIO (test)
44.89BLEU-4
10
Wikibio
79.84Format Adherence Score
8
WikiBio Factual Sentence-level
44.23AUC-PR
8
WikiBio GPT-3.5-Turbo-Instruct (test)
92.5AUC-PR (Nonfactual)
8
WikiBio 22
47.17BLEU
7
WikiBio (test)
84.33RwA
6
WikiBio (val)
2.3Success
4
WikiBio (val)
99.6Fluency
4
WikiBio
—Primary metric
0
WikiBio (test)
—AUC π-Soft-NS
0