List your company

Language Modeling on BILLION WORD (test)

23.5PerplexitySRU++
23.41623.98324.5525.117Feb 24, 2021

Evaluation results

MethodLinks
SRU++Parameters=465M, Effective GPU days=63, k=5, Mixed precision training=true2021.0223.5
TransformerParameters=465M, Effective GPU days=192, Mixed precision training=true2021.0223.9
SRU++Parameters=328M, Effective GPU days=36, Mixed precision training=true2021.0225.1
TransformerEffective GPU days=147, Mixed precision training=true2021.0225.2
TransformerParameters=331M, Effective GPU days=57, Mixed precision training=true2021.0225.6