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

Long-context Language Modeling on LongBench (Generation Score, Attention Recall)

50.6Generation Score

RaBitQCache

1.87614.525527.17539.8245Jun 30, 2026
Updated 24d ago

Evaluation Results

MethodLinks
2026.06
50.690.7
2026.06
50.58100
2026.06
50.4796.9
2026.06
50.4397
2026.06
50.490.7
2026.06
50.3186.9
2026.06
50.2868.4
2026.06
50.1589.8
2026.06
49.2492.9
2026.06
49.0642.7
2026.06
46.5282.6
2026.06
36.7153.3
2026.06
36.2886.1
2026.06
36.2189.8
2026.06
36.1194.7
2026.06
35.9193.1
2026.06
35.8489.3
2026.06
35.78100
2026.06
35.358.1
2026.06
32.5867.4
2026.06
30.527.3
2026.06
21.2847.8
2026.06
9.5220.4
2026.06
3.7510