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

Instruction Following on IFBench (Score %)

43.28IFBench Score

STEP3-VL-10B

15.335222.590129.84537.0999Nov 28, 2025Dec 16, 2025Jan 4, 2026Jan 23, 2026Feb 10, 2026Mar 1, 2026Mar 20, 2026
Updated 26d ago

Evaluation Results

MethodLinks
2026.01
43.28
2026.03
39.7
2026.03
39.7
2026.03
36.9
2026.01
36.65
2026.03
31.6
2026.03
31.3
2026.03
30.6
2026.03
29.9
2026.01
28.14
2026.03
27.9
2026.03
27.9
2026.01
27.47
2026.03
26.6
2026.03
26.2
2026.03
25.9
2026.03
24.8
2026.03
23.8
2026.01
23.47
2026.03
22.1
2025.11
21.27
2025.11
20.7
2025.11
20.56
2025.11
19.75
2025.11
17.72
2025.11
16.86
2025.11
16.41