Ambrosia
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Ambrosia
64.18AUPRC
12
Ambrosia Unambiguous (test)
88.7Recall
11
Ambrosia Ambiguous subset (test)
82.4Recall
11
Ambrosia (out-of-domain)
84.4Single Interpretation Coverage
7
AMBROSIA Unambiguous
94.7Execution Accuracy (EX)
3
AMBROSIA Ambiguous
92Execution Accuracy (EX)
3
Ambrosia Ambiguous
82.4Recall
2
Ambrosia sampled 30 ambiguous examples
0.917Alignment Score
1