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

Multi-needle retrieval on NIAH (M)

90.2Accuracy (NIAH M)

GLM4-9B-Chat-1M

33.72848.38963.0577.711Aug 31, 2024
Updated 1mo ago

Evaluation Results

MethodLinks
2024.08
90.2
2024.08
89.6
2024.08
89
2024.08
84.9
2024.08
83.4
2024.08
82.6
2024.08
82.3
2024.08
82
2024.08
80.1
2024.08
79.5
2024.08
79.2
2024.08
78.9
2024.08
76.2
2024.08
73.7
2024.08
73.2
2024.08
72.7
2024.08
72
2024.08
71.8
2024.08
70
2024.08
68.8
2024.08
68.3
2024.08
65.8
2024.08
65.7
2024.08
64.7
2024.08
64.7
2024.08
64.7
2024.08
64.6
2024.08
60.4
2024.08
58.5
2024.08
53.4
2024.08
52.7
2024.08
43
2024.08
41.9
2024.08
38.8
2024.08
35.9