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

Long-context retrieval and aggregation on RULER 8k (Average Accuracy)

94.24Average Accuracy

NestedKV

11.726433.148254.5775.9918May 26, 2026
Updated 7d ago

Evaluation Results

MethodLinks
2026.05
94.24
2026.05
93.71
2026.05
91.65
2026.05
91.26
2026.05
91.11
2026.05
89.35
2026.05
85.17
2026.05
83.99
2026.05
83.86
2026.05
82.72
2026.05
81.2
2026.05
80.81
2026.05
80.57
2026.05
79.89
2026.05
78.71
2026.05
78.52
2026.05
78.28
2026.05
77.65
2026.05
76.49
2026.05
74.75
2026.05
74.36
2026.05
74.11
2026.05
73.85
2026.05
72.77
2026.05
70.88
2026.05
70.16
2026.05
70.14
2026.05
69.62
2026.05
68.71
2026.05
68.37
2026.05
67.74
2026.05
67.71
2026.05
67.55
2026.05
67.29
2026.05
62.74
2026.05
62.53
2026.05
61.88
2026.05
61.42
2026.05
59.41
2026.05
58.71
2026.05
58
2026.05
56.03
2026.05
55.99
2026.05
55.75
2026.05
55.35
2026.05
54.16
2026.05
53.77
2026.05
52.12
2026.05
51.88
2026.05
50.01
2026.05
49.55
2026.05
49.16
2026.05
47.5
2026.05
45.47
2026.05
42.95
2026.05
41.83
2026.05
40.6
2026.05
38.12
2026.05
38
2026.05
35.42
2026.05
35.15
2026.05
33.99
2026.05
33.28
2026.05
33.27
2026.05
30.95
2026.05
28.42
2026.05
27.76
2026.05
26.34
2026.05
26.22
2026.05
23.85
2026.05
23.14
2026.05
22.53
2026.05
21.99
2026.05
19.79
2026.05
18.52
2026.05
14.9