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

General Instruction Following Accuracy on IFBench

84.1Accuracy

Multi-DPOP

56.33263.54170.7577.959Oct 10, 2025
Updated 22d ago

Evaluation Results

MethodLinks
2025.10
84.1
2025.10
83.9
2025.10
83.5
2025.10
82.6
2025.10
82.4
2025.10
82.1
2025.10
81.3
2025.10
80.7
2025.10
80.5
2025.10
80.2
2025.10
80.1
2025.10
79.8
2025.10
78.6
2025.10
77.9
2025.10
76.8
2025.10
76.5
2025.10
76.3
2025.10
76
2025.10
73.6
2025.10
73.6
2025.10
72.5
2025.10
71.2
2025.10
69.3
2025.10
69.2
2025.10
69
2025.10
68.7
2025.10
66.1
2025.10
65.3
2025.10
64.8
2025.10
57.4