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

MIRB

Benchmarks

Task NameDataset NameSOTA ResultTrend
Multi-image ReasoningMIRB
Accuracy63.57
86
Multi-image UnderstandingMIRB (test)
Accuracy60.8
12
Mathematical Information RetrievalMIRB
MO Duplicate (nDCG@10)85.9
8
Showing 3 of 3 rows