Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Language Understanding on LAMBADA 8 reasoning tasks
Loading...
55.61
Accuracy
CARVE + SWA
50.67
51.9525
53.235
54.5175
Jun 25, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
CARVE + SWA
Model Category=Hybrid,...
2026.06
55.61
GDN-2 + SWA
Model Category=Hybrid,...
2026.06
53.97
CARVE
Model Category=Recurre...
2026.06
53.74
GDN-2
Model Category=Recurre...
2026.06
53.11
Mamba-3 MIMO + SWA
Model Category=Hybrid,...
2026.06
52.72
Mamba-3 SISO + SWA
Model Category=Hybrid,...
2026.06
52.69
KDA + SWA
Model Category=Hybrid,...
2026.06
52.68
Mamba-3 MIMO
Model Category=Recurre...
2026.06
52.39
KDA
Model Category=Recurre...
2026.06
52.28
Gated DeltaNet + SWA
Model Category=Hybrid,...
2026.06
52.25
Gated DeltaNet
Model Category=Recurre...
2026.06
52.07
Mamba-2 + SWA
Model Category=Hybrid,...
2026.06
51.99
Mamba-2
Model Category=Recurre...
2026.06
51.82
Mamba-3 SISO
Model Category=Recurre...
2026.06
51.42
Transformer
Model Category=Hybrid,...
2026.06
50.86
Feedback
Search any
task
Search any
task