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

Web Search Question Answering on SearchQA (held-out test)

87.3Score (%)

SkillOpt

54.5463.04571.5580.055May 22, 2026
Updated 9d ago

Evaluation Results

MethodLinks
2026.05
87.3
2026.05
87.3
2026.05
85.9
2026.05
84.8
2026.05
84.1
2026.05
84
2026.05
83.7
2026.05
83.4
2026.05
83.1
2026.05
83.1
2026.05
82.7
2026.05
82.6
2026.05
82.5
2026.05
82.4
2026.05
82.4
2026.05
82.4
2026.05
81.9
2026.05
81.8
2026.05
81.8
2026.05
81.4
2026.05
81.2
2026.05
80.9
2026.05
80.3
2026.05
80.2
2026.05
80.1
2026.05
79.4
2026.05
79.4
2026.05
79.2
2026.05
78.6
2026.05
77.7
2026.05
77.5
2026.05
77.2
2026.05
76.9
2026.05
76.4
2026.05
76.1
2026.05
75.9
2026.05
75.8
2026.05
75.8
2026.05
75.4
2026.05
74.8
2026.05
74.1
2026.05
73.7
2026.05
73.4
2026.05
73.2
2026.05
72.7
2026.05
72.6
2026.05
71.9
2026.05
71.2
2026.05
69.9
2026.05
68.6
2026.05
68.5
2026.05
68.1
2026.05
66.3
2026.05
66.1
2026.05
65
2026.05
62.4
2026.05
61.4
2026.05
60.7
2026.05
55.8