List your company
ResearchFormerly WizWand

PDE solving on Navier-Stokes 2D (test)

0.0479Relative MSE LossDynFormer
0.0074160.2806830.553950.827217Mar 3, 2026Mar 19, 2026Apr 4, 2026Apr 20, 2026May 6, 2026May 22, 2026Jun 8, 2026
Updated 4mo ago

Evaluation results

MethodLinks
DynFormerModel Variant=Large, Peak GPU memory (GB)=24.6, Number of Parameters=110.7K, Averaged training time (s/epoch)=17.22026.030.0479
DynFormerModel Variant=Medium, Peak GPU memory (GB)=13.3, Number of Parameters=57.9K, Averaged training time (s/epoch)=12.82026.030.0503
DynFormerModel Variant=Tiny, Peak GPU memory (GB)=6.9, Number of Parameters=15.1K, Averaged training time (s/epoch)=12.92026.030.0966
TransolverModel Variant=Large, Peak GPU memory (GB)=23.5, Number of Parameters=95.8K, Averaged training time (s/epoch)=18.62026.030.127
FactFormerModel Variant=Large, Peak GPU memory (GB)=23.7, Number of Parameters=202.6K, Averaged training time (s/epoch)=16.12026.030.141
TransolverModel Variant=Tiny, Peak GPU memory (GB)=8.0, Number of Parameters=22.0K, Averaged training time (s/epoch)=10.92026.030.151
FactFormerModel Variant=Medium, Peak GPU memory (GB)=14.8, Number of Parameters=99.2K, Averaged training time (s/epoch)=13.52026.030.152
FactFormerModel Variant=Tiny, Peak GPU memory (GB)=7.6, Number of Parameters=25.3K, Averaged training time (s/epoch)=12.32026.030.185
TransolverModel Variant=Medium, Peak GPU memory (GB)=13.9, Number of Parameters=42.9K, Averaged training time (s/epoch)=13.82026.030.192
ONOModel Variant=Large, Peak GPU memory (GB)=23.8, Number of Parameters=38.7K, Averaged training time (s/epoch)=21.92026.030.793
BreedStrategy=Breed2026.060.93
OFormerModel Variant=Large, Peak GPU memory (GB)=24.9, Number of Parameters=33.5K, Averaged training time (s/epoch)=17.42026.030.972
ONOModel Variant=Medium, Peak GPU memory (GB)=14.0, Number of Parameters=17.3K, Averaged training time (s/epoch)=16.32026.030.978
OFormerModel Variant=Medium, Peak GPU memory (GB)=13.6, Number of Parameters=11.6K, Averaged training time (s/epoch)=13.82026.030.986
OFormerModel Variant=Tiny, Peak GPU memory (GB)=7.1, Number of Parameters=3.0K, Averaged training time (s/epoch)=12.42026.030.991
GNOTModel Variant=Large, Peak GPU memory (GB)=23.8, Number of Parameters=16.4K, Averaged training time (s/epoch)=28.12026.030.994
GNOTModel Variant=Tiny, Peak GPU memory (GB)=7.5, Number of Parameters=3.8K, Averaged training time (s/epoch)=11.12026.030.996
GNOTModel Variant=Medium, Peak GPU memory (GB)=14.0, Number of Parameters=7.3K, Averaged training time (s/epoch)=16.82026.030.999
ONOModel Variant=Tiny, Peak GPU memory (GB)=8.2, Number of Parameters=8.5K, Averaged training time (s/epoch)=13.22026.030.999
SBALStrategy=SBAL2026.061.02
OGAS-UStrategy=OGAS-U2026.061.03
SobolStrategy=Sobol2026.061.04
OGAS-LStrategy=OGAS-L2026.061.05
Top-KStrategy=Top-K2026.061.06