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

ARO

Benchmarks

Task NameDataset NameSOTA ResultTrend
Multimodal Compositional UnderstandingARO
Relational Score86.7
27
Compositional ReasoningARO
Relation Score83.6
17
Vision-Language Compositional ReasoningARO
Accuracy0.804
14
Compositional EvaluationARO-A (test)
Accuracy84.49
13
Order SensitivityARO
Flickr30K Order99.4
13
Relational UnderstandingARO
Relation Accuracy59
6
Compositional AttributionARO Attribution
Accuracy77.1
4
Compositional RetrievalARO Retrieval
Accuracy83.6
4
Compositional ReasoningARO (test)
Relation Score83.7
4
Vision-Language Compositional ReasoningARO (test)
VG-Rel71.4
4
Showing 10 of 10 rows