List your company

Machine Translation on WMT English-German (EN-DE) 2014 (newstest2014)

30.77BLEURK2-block (learnable gamma_i)
24.353226.019127.68529.3509May 7, 2021Jun 28, 2021Aug 19, 2021Oct 11, 2021Dec 2, 2021Jan 23, 2022Mar 17, 2022

Evaluation results

MethodLinks
RK2-block (learnable gamma_i)Layers=12-6, #Param=286M, Steps=100K2022.0330.7729.6
BERT-fused modelLayers=6-62022.0330.75—
RK2-blockLayers=12-6, #Param=286M, Steps=100K2022.0330.5829.4
Multiscale CollaborativeLayers=18-6, #Param=512M, Steps=300K2022.0330.56—
RK4-blockLayers=12-6, #Param=286M, Steps=100K2022.0330.5529.4
RK2-block (learnable gamma_i)Layers=6-6, #Param=211M, Steps=100K2022.0330.5329.4
RK4-blockLayers=6-6, #Param=211M, Steps=100K2022.0330.3929.3
RK2-block (learnable gamma_i)Layers=24-6, #Param=118M, Steps=50K2022.0330.2929.2
SDTLayers=48-6, #Param=192M, Steps=50K2022.0330.2129
MacaronNetLayers=6-62022.0330.2—
RK2-blockLayers=6-6, #Param=211M, Steps=100K2022.0330.1129
ADMINLayers=60-12, #Param=262M, Steps=250K2022.0330.0129.5
Depth growingLayers=8-8, #Param=270M, Steps=800K2022.0329.92—
Residual-block-BigLayers=12-6, #Param=286M, Steps=100K2022.0329.9128.9
RK2-blockLayers=24-6, #Param=118M, Steps=50K2022.0329.8528.7
RK4-blockLayers=24-6, #Param=118M, Steps=50K2022.0329.828.8
Residual-blockLayers=24-6, #Param=118M, Steps=50K2022.0329.4328.3
Transformer-DLCLLayers=30-6, #Param=137M, Steps=50K2022.0329.328.6
Residual-block-BigLayers=6-6, #Param=211M, Steps=100K2022.0329.2128.1
RK4-blockLayers=6-6, #Param=61M, Steps=50K2022.0329.0327.9
RK2-block (learnable gamma_i)Layers=6-6, #Param=61M, Steps=50K2022.0328.8927.7
RK2-blockLayers=6-6, #Param=61M, Steps=50K2022.0328.6727.5
TransformerLayers=6-6, #Param=213M, Steps=100K2022.0328.4—
Residual-blockLayers=6-6, #Param=61M, Steps=50K2022.0327.8926.8
Transformer (base)2021.0527.3—
ResMLP-12layers=12, dimension=512, hidden MLP size=2,0482021.0526.8—
ResMLP-6layers=6, dimension=512, hidden MLP size=2,0482021.0526.4—
ConvS2S2021.0525.2—
GNMT2021.0524.6—