PDE solving on Kuramoto-Sivashinsky 1D (test)
0.0086Relative MSE LossDynFormer
Evaluation results
| Method | Links | |
|---|---|---|
| DynFormerModel Variant=Medium, Peak GPU memory (GB)=2.8, Number of Parameters=5.9K, Averaged training time (s/epoch)=5.72026.03 | 0.0086 | |
| DynFormerModel Variant=Large, Peak GPU memory (GB)=12.9, Number of Parameters=163.5K, Averaged training time (s/epoch)=11.42026.03 | 0.023 | |
| DynFormerModel Variant=Tiny, Peak GPU memory (GB)=1.0, Number of Parameters=1.2K, Averaged training time (s/epoch)=3.92026.03 | 0.169 | |
| OFormerModel Variant=Large, Peak GPU memory (GB)=11.3, Number of Parameters=1.4M, Averaged training time (s/epoch)=5.32026.03 | 0.177 | |
| TransolverModel Variant=Large, Peak GPU memory (GB)=13.5, Number of Parameters=2.8M, Averaged training time (s/epoch)=7.22026.03 | 0.193 | |
| FactFormerModel Variant=Large, Peak GPU memory (GB)=12.3, Number of Parameters=638.0K, Averaged training time (s/epoch)=6.82026.03 | 0.338 | |
| FactFormerModel Variant=Medium, Peak GPU memory (GB)=2.9, Number of Parameters=51.4K, Averaged training time (s/epoch)=3.62026.03 | 0.374 | |
| TransolverModel Variant=Tiny, Peak GPU memory (GB)=0.8, Number of Parameters=80.4K, Averaged training time (s/epoch)=2.02026.03 | 0.466 | |
| TransolverModel Variant=Medium, Peak GPU memory (GB)=2.7, Number of Parameters=369.7K, Averaged training time (s/epoch)=3.72026.03 | 0.506 | |
| FactFormerModel Variant=Tiny, Peak GPU memory (GB)=0.8, Number of Parameters=6.6K, Averaged training time (s/epoch)=2.02026.03 | 0.512 | |
| OFormerModel Variant=Medium, Peak GPU memory (GB)=2.7, Number of Parameters=92.0K, Averaged training time (s/epoch)=3.12026.03 | 0.647 | |
| ONOModel Variant=Tiny, Peak GPU memory (GB)=0.8, Number of Parameters=21.1K, Averaged training time (s/epoch)=2.32026.03 | 0.771 | |
| ONOModel Variant=Medium, Peak GPU memory (GB)=3.0, Number of Parameters=152.1K, Averaged training time (s/epoch)=5.12026.03 | 0.789 | |
| OFormerModel Variant=Tiny, Peak GPU memory (GB)=0.9, Number of Parameters=11.6K, Averaged training time (s/epoch)=2.22026.03 | 0.949 | |
| GNOTModel Variant=Tiny, Peak GPU memory (GB)=0.9, Number of Parameters=14.1K, Averaged training time (s/epoch)=2.52026.03 | 1 | |
| GNOTModel Variant=Medium, Peak GPU memory (GB)=3.0, Number of Parameters=59.2K, Averaged training time (s/epoch)=4.72026.03 | 1 | |
| GNOTModel Variant=Large, Peak GPU memory (GB)=14.6, Number of Parameters=993.3K, Averaged training time (s/epoch)=9.02026.03 | 1 | |
| ONOModel Variant=Large, Peak GPU memory (GB)=13.2, Number of Parameters=1.2M, Averaged training time (s/epoch)=10.82026.03 | 1.05 |