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

Long-context understanding on RULER 32K

94.48Accuracy

Qwen2.5-14B

54.765.027575.35585.6825Nov 18, 2025Nov 22, 2025Nov 26, 2025Nov 30, 2025Dec 4, 2025Dec 8, 2025Dec 12, 2025
Updated 19d ago

Evaluation Results

MethodLinks
2025.11
94.48
2025.11
94.41
2025.11
94.22
2025.11
94.13
2025.12
92.33
2025.12
92.33
2025.11
92.19
2025.12
92.11
2025.12
92.08
2025.12
92.07
2025.12
91.9
2025.12
91.9
2025.12
91.9
2025.12
91.81
2025.12
91.79
2025.12
91.74
2025.12
91.67
2025.12
91.67
2025.12
91.55
2025.12
87.89
2025.11
80.94
2025.11
80.8
2025.11
80.7
2025.11
80.64
2025.11
76.76
2025.12
56.23