Autoregressive Super-resolution on PF100 50 → 100 Cahn-Hilliard phase-field (test)
0.32RL2DM-UNO
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DM-UNOEnhancement Method=Diffusion, Backbone=UNO, Size (NO + Decoder)=6.38M + 1.44M, Inference Time per Frame (s)=0.178 ± 0.0212026.04 | 0.32 | 0.76 | 2.4 | |
| DM-FNOEnhancement Method=Diffusion, Backbone=FNO, Size (NO + Decoder)=6.33M + 1.44M, Inference Time per Frame (s)=0.176 ± 0.0222026.04 | 0.3 | 0.77 | 2.28 | |
| FNOEnhancement Method=None, Backbone=FNO2026.04 | 0.26 | 0.85 | 1.99 | |
| MENO-UNOEnhancement Method=MeanFlow, Backbone=UNO, Size (NO + Decoder)=6.38M + 1.44M, Inference Time per Frame (s)=0.031 ± 0.0212026.04 | 0.13 | 0.94 | 9.89 | |
| MENO-FNOEnhancement Method=MeanFlow, Backbone=FNO, Size (NO + Decoder)=6.33M + 1.44M, Inference Time per Frame (s)=0.028 ± 0.0222026.04 | 0.08 | 0.96 | 6.52 |