List your company

Language Modeling on WikiText-2 32768 context length (val)

2.048LossNTK+affine
2.039242.098372.15752.21663Jun 3, 2026

Evaluation results

MethodLinks
NTK+affineTrain context length=1024, Eval context length=327682026.062.048
YaRN+affineTrain context length=1024, Eval context length=327682026.062.267