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

OOD

Benchmarks

Task NameDataset NameSOTA ResultTrend
Speculative decoding evaluationOOD Mean
Speedup5.21
20
Unsupervised Object SegmentationOOD 1.0 (test)
FG-ARI7,824
16
OOD DetectionOOD
AUC (Confidence)0.822
9
Speculative DecodingOOD
Block Efficiency2.13
5
Defective Dialog DetectionOOD Shopping n = 105 (test)
Precision48
5
Unsupervised image annotationOOD set
NMI0.54
5
Referential CommunicationOOD set
Accuracy92.7
5
Open-ended DialogueOOD Average
Win Rate60.5
4
Table UnderstandingOOD Table S2 (test)
ROUGE-L40.38
4
Table UnderstandingOOD Table S1 (test)
Accuracy80.2
4
Showing 10 of 10 rows