Wizard of Wikipedia
Benchmarks
TaskDatasetSOTA resultTrendResults
Wizard of Wikipedia (WoW)
78.6Performance Score
32
Wizard of Wikipedia (WoW)
97.7Win Rate
20
Wizard of Wikipedia (WoW) (dev)
16.4F1 Score
19
Wizard of Wikipedia (WoW) seen (test)
27.29BLEU-1
13
Wizard of Wikipedia unseen (test)
27.68BLEU-1
11
WoW (Wizard of Wikipedia) unseen (test)
20.7ROUGE-1
10
Wizard of Wikipedia (WoW) Seen (test)
21.7ROUGE-1
10
Wizard-of-Wikipedia (WoW) (test)
100BLEU
9
WoW (Wizard of Wikipedia) (test)
11.38F1 Score
8
Wizard-of-Wikipedia KILT (test)
15.78F1 Score
8
WoW (Wizard of Wikipedia)
63.43Entity in Context
7
Wizard of Wikipedia (WW)
12.4Perplexity
4
Wizard of Wikipedia (WoW) (test seen)
41.34Relevance Win Rate
1