List your company
ResearchFormerly WizWand

PDE Solving on Darcy-Flow 2d (test)

0.0001Relative MSEFactFormer
-0.039880.2298290.4995380.769248Oct 9, 2024Jan 2, 2025Mar 28, 2025Jun 21, 2025Sep 14, 2025Dec 8, 2025Mar 3, 2026
Updated 8mo ago

Evaluation results

MethodLinks
FactFormerModel Variant=Large, Peak GPU memory (GB)=13.0, Number of Parameters=7.3M, Averaged training time (s/epoch)=4.82026.030.0001
FactFormerModel Variant=Medium, Peak GPU memory (GB)=2.9, Number of Parameters=476.1K, Averaged training time (s/epoch)=1.02026.030.0001
DynFormerModel Variant=Large, Peak GPU memory (GB)=14.1, Number of Parameters=3.4M, Averaged training time (s/epoch)=3.62026.030.0002
DynFormerModel Variant=Medium, Peak GPU memory (GB)=2.4, Number of Parameters=162.9K, Averaged training time (s/epoch)=1.42026.030.0003
DynFormerModel Variant=Tiny, Peak GPU memory (GB)=0.7, Number of Parameters=32.6K, Averaged training time (s/epoch)=0.72026.030.0004
FactFormerModel Variant=Tiny, Peak GPU memory (GB)=0.7, Number of Parameters=38.9K, Averaged training time (s/epoch)=0.42026.030.0008
TransolverModel Variant=Medium, Peak GPU memory (GB)=2.6, Number of Parameters=283.6K, Averaged training time (s/epoch)=1.02026.030.001
OFormerModel Variant=Large, Peak GPU memory (GB)=12.3, Number of Parameters=1.7M, Averaged training time (s/epoch)=4.52026.030.001
TransolverModel Variant=Tiny, Peak GPU memory (GB)=0.8, Number of Parameters=65.1K, Averaged training time (s/epoch)=0.42026.030.0013
OFormerModel Variant=Medium, Peak GPU memory (GB)=2.6, Number of Parameters=91.9K, Averaged training time (s/epoch)=0.92026.030.0031
OFormerModel Variant=Tiny, Peak GPU memory (GB)=0.9, Number of Parameters=11.6K, Averaged training time (s/epoch)=0.42026.030.0045
TransolverModel Variant=Large, Peak GPU memory (GB)=12.3, Number of Parameters=2.8M, Averaged training time (s/epoch)=4.62026.030.0069
ONOModel Variant=Medium, Peak GPU memory (GB)=3.1, Number of Parameters=151.0K, Averaged training time (s/epoch)=1.22026.030.0094
ONOModel Variant=Tiny, Peak GPU memory (GB)=0.9, Number of Parameters=25.6K, Averaged training time (s/epoch)=0.52026.030.0095
OursTraining Paradigm=Hybrid2024.100.0162
Neural SolverLearning Paradigm=Neural Solver2024.100.0187
MLP + basisLearning Paradigm=Supervised2024.100.0356
MLP + basisTraining Paradigm=Supervised2024.100.0378
GNOTModel Variant=Tiny, Peak GPU memory (GB)=0.8, Number of Parameters=8.0K, Averaged training time (s/epoch)=0.52026.030.076
GNOTModel Variant=Medium, Peak GPU memory (GB)=3.1, Number of Parameters=103.5K, Averaged training time (s/epoch)=1.22026.030.076
GNOTModel Variant=Large, Peak GPU memory (GB)=11.6, Number of Parameters=786.1K, Averaged training time (s/epoch)=4.12026.030.0762
PINOLearning Paradigm=Hybrid2024.100.101
PINOTraining Paradigm=Hybrid2024.100.117
PI-DeepONetTraining Paradigm=Hybrid2024.100.272
PI-DeepONetLearning Paradigm=Hybrid2024.100.276
ONOModel Variant=Large, Peak GPU memory (GB)=11.3, Number of Parameters=1.2M, Averaged training time (s/epoch)=4.02026.030.337
PO-DeepONetTraining Paradigm=Unsupervised2024.100.832
PO-DeepONetLearning Paradigm=Unsupervised2024.100.833
PINNS-multi-optLearning Paradigm=Unsupervised2024.100.838
P2INNsLearning Paradigm=Unsupervised2024.100.838
PPINNsLearning Paradigm=Unsupervised2024.100.847
PINNs+L-BFGSLearning Paradigm=Unsupervised2024.100.999
PPINNsTraining Paradigm=Unsupervised2024.100.999