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

Long-context retrieval and aggregation on RULER 16k

93.07Average Accuracy

Full KV

11.357232.571153.78574.9989May 26, 2026
Updated 7d ago

Evaluation Results

MethodLinks
2026.05
93.07
2026.05
91.46
2026.05
91.13
2026.05
90.44
2026.05
88.36
2026.05
88.1
2026.05
84.65
2026.05
84.04
2026.05
82.67
2026.05
82.19
2026.05
81.68
2026.05
81.05
2026.05
79.89
2026.05
79.3
2026.05
77.8
2026.05
77.05
2026.05
76.84
2026.05
76.83
2026.05
76.6
2026.05
76.54
2026.05
74.73
2026.05
73.57
2026.05
73.47
2026.05
73
2026.05
72.51
2026.05
70
2026.05
69.19
2026.05
69.13
2026.05
68.84
2026.05
66.99
2026.05
66.63
2026.05
66.49
2026.05
66.46
2026.05
65.26
2026.05
65.22
2026.05
63.34
2026.05
62.48
2026.05
62.2
2026.05
61.73
2026.05
60.79
2026.05
60
2026.05
55.04
2026.05
54.55
2026.05
54.2
2026.05
53.87
2026.05
52.38
2026.05
52.15
2026.05
50.56
2026.05
50.2
2026.05
49.75
2026.05
49.74
2026.05
49.47
2026.05
46.9
2026.05
46.82
2026.05
45.91
2026.05
45.28
2026.05
44.92
2026.05
39.64
2026.05
37.95
2026.05
37.47
2026.05
37.41
2026.05
35.85
2026.05
35.81
2026.05
34.98
2026.05
34.2
2026.05
31.25
2026.05
30.81
2026.05
30.44
2026.05
29.76
2026.05
29.51
2026.05
27.23
2026.05
24.23
2026.05
22.72
2026.05
22.48
2026.05
16.8
2026.05
14.5