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

Abductive Commonsense Reasoning on aNLI (test)

92.9Accuracy

Human Performance

48.28459.86771.4583.033Mar 4, 2022Nov 2, 2022Jul 4, 2023Mar 4, 2024Nov 2, 2024Jul 4, 2025Mar 5, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2022.10
92.9
2022.10
92.8
2022.10
92.2
2026.03
91.4
2026.03
89
2026.03
85.6
2022.03
85.51
2022.03
85.51
2026.03
83.4
2026.03
83.3
2022.03
82.77
2022.03
82.77
2022.03
82.35
2022.03
82.35
2026.03
82.2
2026.03
81.2
2022.03
80.74
2022.03
80.74
2026.03
79.6
2022.03
79.5
2022.03
79.5
2022.03
77.8
2022.03
77.12
2022.03
77.12
2026.03
76
2026.03
75
2026.03
74.7
2026.03
73.8
2026.03
73.2
2026.03
73.2
2026.03
72.7
2026.03
72.5
2026.03
72.5
2026.03
71.2
2026.03
70.8
2022.03
66.75
2026.03
65.6
2022.03
65.27
2022.03
63.18
2022.03
63.04
2022.03
61.88
2026.03
61.8
2026.03
61.7
2026.03
61.5
2022.03
61.1
2026.03
59.9
2026.03
59.2
2026.03
56.5
2026.03
55.9
2026.03
55.2
2026.03
54.8
2026.03
51
2022.03
50