List your company
ResearchFormerly WizWand

Multi-modal Large Language Model Evaluation on MME (test)

1,515.4MME ScoreOPERA
1,470.161,481.9051,493.651,505.395Nov 29, 2023
Updated 8mo ago

Evaluation results

MethodLinks
OPERADecoding Strategy=OPERA, Model=LLaVA-1.5 7B2023.111,515.4
GreedyDecoding Strategy=Greedy, Model=LLaVA-1.5 7B2023.111,510.7
BeamDecoding Strategy=Beam Search, Model=LLaVA-1.5 7B2023.111,504.3
DoLaDecoding Strategy=DoLa, Model=LLaVA-1.5 7B2023.111,480.1
NucleusDecoding Strategy=Nucleus Sampling, Model=LLaVA-1.5 7B2023.111,471.9