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

Snips

Benchmarks

Task NameDataset NameSOTA ResultTrend
Intent ClassificationSNIPS (test)
Accuracy99.7
40
Natural Language UnderstandingSnips (test)
Intent Acc99.29
27
Slot FillingSNIPS (test)
F1 Score0.983
25
Intent DetectionSNIPS Type 3 (test)
Accuracy96.2
15
Intent DetectionSNIPS Type 2 (test)
Accuracy98.9
15
Intent DetectionSNIPS Type 1 (test)
Accuracy98.6
15
Spoken Language UnderstandingSNIPS
Slot F197
15
Unknown Intent DetectionSNIPS (test)
Macro F184.1
15
Slot fillingSNIPS all target domains
F1 Score0.578
12
Slot FillingSNIPS SearchCreativeWork zero-shot (test)
F1-score72.88
11
Slot FillingSNIPS RateBook zero-shot (test)
F1 Score47.53
11
Slot FillingSNIPS PlayMusic zero-shot (test)
F1 Score66.42
11
Slot FillingSNIPS zero-shot Average
F1 Score61.07
11
Slot FillingSNIPS zero-shot (SearchScreeningEvent)
F1 Score51.42
11
Slot FillingSNIPS GetWeather zero-shot
F1 Score65.36
11
Slot FillingSNIPS BookRestaurant zero-shot
F1 Score63.77
11
Slot FillingSNIPS zero-shot AddToPlaylist
F1 Score68.7
11
ClusteringSNIPS (test)
NMI89.3
11
OOD Intent DetectionSNIPS standard (test)
Macro F192.32
10
Intent ClassificationSNIPS (unsupervised)
Accuracy97
9
Intent DetectionSNIPS
F1 Score98.7
8
Slot FillingSNIPS 5-shot
We Score89.39
8
Slot FillingSNIPS
CT10
7
Open intent recognitionSNIPS
Accuracy96.1
6
Slot FillingSNIPS 1-shot
Average Score70.44
6
Showing 25 of 38 rows