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

SLURP

Benchmarks

Task NameDataset NameSOTA ResultTrend
Selective capability unlearningSLURP Retain Set 1.0 (DR)
IF90.18
40
Slot FillingSLURP (test)
F1 Score80.61
26
Intent ClassificationSLURP (test)
Accuracy (IC)90.07
26
Selective capability unlearningSLURP Forget Set (DF) 1.0
IF95.27
20
Intent DetectionSLURP
Accuracy89.16
16
ASR RescoringSLURP (test)
WER23.77
11
Spoken Language UnderstandingSLURP 6 tasks
Avg Accuracy84.41
9
Spoken Language UnderstandingSLURP 3 tasks
Average Accuracy84.41
9
Spoken Language UnderstandingSLURP Oracle
Intent Accuracy91.57
8
Slot FillingSLURP Reference transcriptions (test)
SLU F1 Score86.3
6
Slot FillingSLURP Whisper Medium transcriptions (test)
SLU F1 Score66.5
6
Slot FillingSLURP Whisper Small transcriptions (test)
SLU-F164.6
6
Slot FillingSLURP Whisper Base transcriptions WER 26.4% (test)
SLU F1 Score59.1
6
Slot FillingSLURP Whisper Tiny transcriptions (test)
SLU F1 Score56.4
6
Spoken Language UnderstandingSLURP (test)
Scenario Accuracy78.76
5
Spoken Language UnderstandingSLURP 0 dB
Intent Accuracy33.29
4
Spoken Language UnderstandingSLURP 2.5 dB
Intent Accuracy46.68
4
Spoken Language UnderstandingSLURP 5 dB
Intent Accuracy59.47
4
Spoken Language UnderstandingSLURP 10 dB
Intent Accuracy74.94
4
Spoken Language UnderstandingSLURP Clean
Intent Accuracy86.02
4
Speech Intent Classification and Slot FillingSLURP (test)
Intent Accuracy90.68
4
Showing 21 of 21 rows