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

Relation Hallucination Evaluation on Reefknot 1.0 (full)

46.7Perception Y/N Accuracy

MiniGPT4-v2

31.339235.327139.31543.3029Aug 18, 2024
Updated 1mo ago

Evaluation Results

MethodLinks
2024.08
46.77861.343.7368.565.8839.88
2024.08
42.2156.772.4733.5321.8873.0152.55
2024.08
40.759.3548.9334.1929.1954.4557.47
2024.08
39.8857.0750.9833.9721.3549.4560.3
2024.08
37.6768.0552.9333.9951.0454.5651.41
2024.08
37.5856.3367.0732.2223.659.3456.39
2024.08
37.2347.9570.1429.8918.5466.1857.1
2024.08
34.0950.4758.0927.0816.8756.4762.03
2024.08
33.5647.5371.0233.1616.3374.9655.81
2024.08
32.7944.1957.6733.7514.8552.7262.61
2024.08
32.5640.9342.726.2711.5348.7868.32
2024.08
31.9348.6547.6327.6516.7145.9665.73