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

Long-context Language Modeling on RULER (4K Context)

95.3Accuracy

Full KV

16.36436.85757.3577.843Oct 9, 2025
Updated 1d ago

Evaluation Results

MethodLinks
2025.10
95.3
2025.10
82
2025.10
81.3
2025.10
76.7
2025.10
76
2025.10
75.6
2025.10
75.4
2025.10
75.3
2025.10
74.5
2025.10
68.6
2025.10
67.8
2025.10
67.6
2025.10
67.5
2025.10
67.1
2025.10
65.9
2025.10
65.5
2025.10
61.8
2025.10
59.9
2025.10
57.5
2025.10
57.1
2025.10
55.2
2025.10
54.2
2025.10
47.7
2025.10
46.3
2025.10
37.7
2025.10
33.5
2025.10
24.1
2025.10
21.5
2025.10
19.4