LongLaMP
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
LongLaMP
42.5R1
32
LongLaMP Wri
30.79R1 Score
16
LongLaMP Rev
34.81R1 Score
16
LongLaMP Abs
43.91ROUGE-1
16
LongLaMP
30.8R1
16
LongLaMP
23.07ROUGE-L
12
LongLaMP
27.84ROUGE-L
12
LongLaMP
38.92ROUGE-L
12
LONGLAMP Topic Writing (user-based)
0.314ROUGE-1
9
LONGLAMP Abstract Generation (user-based split)
37.16ROUGE-1
9
LONGLAMP Product Review (user-based split)
36.63ROUGE-1
9
LongLaMP Pair A Writing (test)
30.79ROUGE-1
8
LongLaMP (Pair A) - Review (test)
33.03ROUGE-1
8
LongLaMP (Pair A) - Abstract (test)
41.35ROUGE-1
8
LONGLAMP-2 OOD
28.7ROUGE-1 Score
7
LongLaMP-3 (random)
266ROUGE-1
7
LongLaMP-2 (random)
29.2ROUGE-1 Score
7
LongLaMP-1 (random)
34.1R-1
7
LongLaMP
74Alignment Score
7
LongLaMP PTW Qwen3-4B (test)
21.02ROUGE-L
6
LongLaMP PRW Qwen3-4B (test)
26.68ROUGE-L
6
LongLaMP PAG Qwen3-4B (test)
37.27ROUGE-L
6
LongLaMP PTW (test)
22.18ROUGE-L
6
LongLaMP PRW (test)
27.71ROUGE-L
6
LongLaMP PAG (test)
37.56ROUGE-L
6