PDEBench fluid evolution
Neural surrogates that roll out PDE solutions with less error (placeholder)
README.md
| issue | 3 |
|---|---|
| status | done |
| tags | training |
pushforward-training
Push-forward training on unrolled steps.
What happened
- 2026-10-02: 0.0731 nRMSE (−0.0179 on the baseline)
- 2026-10-08: 0.0745 nRMSE (−0.0165 on the baseline)
- 2026-10-14: 0.0684 nRMSE (−0.0226 on the baseline)
Next
Bring the spectral filter in.