Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Ambrosia

Benchmarks

Task NameDataset NameSOTA ResultTrend
Text-to-SQLAmbrosia
AUPRC64.18
12
Text-to-SQL ParsingAmbrosia Unambiguous (test)
Recall88.7
11
Text-to-SQL ParsingAmbrosia Ambiguous subset (test)
Recall82.4
11
Text-to-SQLAmbrosia (out-of-domain)
Single Interpretation Coverage84.4
7
Text-to-SQL ambiguity resolutionAmbrosia Ambiguous
Recall82.4
2
Interpretation AlignmentAmbrosia sampled 30 ambiguous examples
Alignment Score0.917
1
Showing 6 of 6 rows