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

Multi-modal Hallucination Evaluation on MMHal-Bench v1.0 (test)

2.14Overall Score

InstructBLIP

1.361.56251.7651.9675Dec 12, 2023
Updated 1mo ago

Evaluation Results

MethodLinks
2023.12
2.14582.751.751.252.082.54.081.51.17
2023.12
2.13502.952.152.291.971.531.982.022.19
2023.12
2.1583.422.081.331.922.173.671.171.08
2023.12
2.08622.942.012.271.642.352.141.671.63
2023.12
2.08522.7522.332.081.51.911.912.16
2023.12
2.05612.331.2522.51.53.332.331.17
2023.12
2.05682.921.832.421.922.252.251.751.08
2023.12
1.89641.580.752.751.831.832.52.171.67
2023.12
1.8651.221.852.231.742.132.481.031.58
2023.12
1.696820.251.421.671.672.672.51.33
2023.12
1.55761.3301.831.1722.581.671.83
2023.12
1.39710.751.832.160.911.251.330.911.91