List your company

Language Modeling on OpenWebText2 (val)

17.12PerplexityDenseFormer
17.060417.462717.86518.2673Feb 4, 2024

Evaluation results

MethodLinks
DenseFormerDilation × DWA Period=1 × 1, Depth=72, Parameters # (M)=548.362024.0217.122.93
DenseFormerDilation × DWA Period=4 × 1, Depth=72, Parameters # (M)=548.352024.0217.173.6
DenseFormerDilation × DWA Period=4 × 5, Depth=72, Parameters # (M)=548.352024.0217.213.9
TransformerDepth=90, Parameters # (M)=675.782024.0217.443.32
TransformerDepth=84, Parameters # (M)=633.312024.0217.483.54
TransformerDepth=72, Parameters # (M)=548.352024.0217.824.08
DenseFormerDilation × DWA Period=1 × 1, Depth=48, Parameters # (M)=378.452024.0217.844.65
DenseFormerDilation × DWA Period=4 × 1, Depth=48, Parameters # (M)=378.452024.0217.865.31
DenseFormerDilation × DWA Period=4 × 5, Depth=48, Parameters # (M)=378.452024.0217.875.72
TransformerDepth=64, Parameters # (M)=491.722024.0217.944.57
Skips With GainsDepth=48, Parameters # (M)=378.452024.0218.455.72
TransformerDepth=48, Parameters # (M)=378.452024.0218.615.94