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

Discriminative Hallucination Evaluation on AMBER (test)

86.8Accuracy

RC-DPO

69.567274.041178.51582.9889May 27, 2026Jun 2, 2026Jun 9, 2026Jun 15, 2026Jun 22, 2026Jun 28, 2026Jul 5, 2026
Updated 18d ago

Evaluation Results

MethodLinks
2026.05
86.889.5690.0789.81
2026.05
86.7389.190.5489.81
2026.05
86.4689.689.4189.5
2026.05
86.4289.1389.9589.53
2026.05
86.3888.9790.0989.52
2026.05
86.3487.6291.8389.67
2026.05
86.2387.2892.1189.63
2026.05
86.2287.691.6389.57
2026.05
86.0287.6191.1389.33
2026.05
85.9987.391.6589.41
2026.05
85.7889.488.4988.94
2026.05
85.6286.4492.2189.22
2026.05
85.2887.5290.0588.76
2026.07
83.7--87.8
2026.07
83.6--87.7
2026.07
83.5--87.7
2026.07
83.3--87.7
2026.07
83.1--87.5
2026.07
82.6--87.1
2026.05
77.3686.7676.6481.38
2026.07
76.8--79.7
2026.07
76.5--79.3
2026.07
76.4--78.8
2026.07
76.3--78.8
2026.07
76.1--79.1
2026.07
75.9--78.4
2026.07
73.4--75.1
2026.07
72.7--75.9
2026.07
72.7--75.1
2026.07
72.6--75.7
2026.05
72.486.0968.3276.18
2026.07
72.3--75.3
2026.07
72.2--74.8
2026.07
72--74.9
2026.07
71.5--74.1
2026.07
71.5--74.2
2026.07
71.5--74.4
2026.07
71.4--74.3
2026.07
70.9--74.1
2026.05
70.6986.0365.2174.18
2026.05
70.3985.4665.2674
2026.05
70.2385.2865.1673.87