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

APIGen

Benchmarks

Task NameDataset NameSOTA ResultTrend
Tool RetrievalAPIGen
NDCG@100.8575
44
Language Model CalibrationAPIGen (test)
smECE1.7
21
Confidence EstimationAPIGen (test)
smECE0.02
21
Argument GenerationAPIGen sampled (test)
Argument F1 (2 calls)88.1
15
Tool SelectionAPIGen sampled (test)
Tool Selection F1 (2 calls)99.4
15
Tool-callingAPIGen (test)
smECE0.02
9
Tool-Calling and Answer GenerationAPIGen-MT (test)
Action Recall90.18
4
Function CallingAPIGen (test)
Score (Single)89.6
2
Showing 8 of 8 rows