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

Hallucination Mitigation on HallusionBench visual-dependent setting (test)

33.21qAcc (Question Accuracy)

ReVisiT

9.560415.700221.8427.9798Jun 11, 2025
Updated 20d ago

Evaluation Results

MethodLinks
2025.06
33.2139.1359.3966.0752.19
2025.06
32.8536.0961.7664.2656.93
2025.06
32.4937.3959.965.751.82
2025.06
32.1336.0961.5965.755.11
2025.06
31.0536.0958.0461.7352.19
2025.06
30.3233.4860.0762.8254.38
2025.06
29.9634.7861.0865.753.65
2025.06
29.635.2261.2564.6254.74
2025.06
28.8831.7457.5357.7654.74
2025.06
28.8835.6557.3662.8249.64
2025.06
28.5236.5256.6862.0949.27
2025.06
28.1633.9157.0260.6550.73
2025.06
27.0830.8758.0462.0951.46
2025.06
25.9933.9156.3561.7348.54
2025.06
20.2226.5251.4445.1354.38
2025.06
14.4422.6147.5544.7747.45
2025.06
13.3622.6149.2445.8549.27
2025.06
12.2721.348.2244.7748.18
2025.06
11.9123.0448.2243.6849.27
2025.06
11.5522.6148.2244.448.54
2025.06
10.4720.4346.3641.5247.81