---
license: cc-by-nc-sa-4.0
---

We take the dataset from [DiffusionPDE](https://github.com/jhhuangchloe/DiffusionPDE). For convenience, we provide our processed version on Hugging Face (see Appendix D & E in [our FunDPS paper](https://arxiv.org/abs/2505.17004) for details). The processing scripts are provided under `utils/` [here](https://github.com/neuraloperator/FunDPS). It is worth noting that we normalized the datasets to zero mean and 0.5 standard deviation to follow [EDM](https://github.com/NVlabs/edm)'s practice.