Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Language Modeling on AdaptEval (test)

1.6598NLL

Layer-wise Dynamic TTA (SCALENET)

1.2626283.9435396.624459.305361Feb 10, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.02
1.6598
2026.02
1.6692
2026.02
1.6741
2026.02
1.679
2026.02
1.7048
2026.02
1.8739
2026.02
1.8777
2026.02
1.8889
2026.02
1.9168
2026.02
2.0829
2026.02
2.0842
2026.02
2.0892
2026.02
2.0925
2026.02
2.1107
2026.02
2.1303
2026.02
2.1399
2026.02
2.1665
2026.02
2.1682
2026.02
2.1725
2026.02
2.1757
2026.02
2.1784
2026.02
2.1805
2026.02
2.1809
2026.02
2.185
2026.02
2.1907
2026.02
2.1997
2026.02
2.2114
2026.02
2.2657
2026.02
5.8488
2026.02
9.6803
2026.02
11.497
2026.02
11.5891