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

WebText

Benchmarks

Task NameDataset NameSOTA ResultTrend
Language ModelingWebText
Mauve58
33
Text GenerationWebText
ROUGE-137.5
15
Language ModelingWebText (test)
Diversity (Div)0.87
14
Language GenerationWebText (completions)
Perplexity (PPL)10.16
7
Text Generation Evaluation CorrelationWebText (test)
Perplexity (PPL)0.643
3
Open-ended Text GenerationWebText (test)
Same Preference Count97
2
Showing 6 of 6 rows