---
title: SirenFNO-Darcy32
canonical_url: "https://www.modelscope.cn/datasets/OneScience/SirenFNO-Darcy32"
md_url: "https://www.modelscope.cn/datasets/OneScience/SirenFNO-Darcy32.md"
repository: OneScience/SirenFNO-Darcy32
last_updated: 2026-11-19
license: other
storage_size: "188 MB"
downloads: 1
stars: 0
---

# SirenFNO-Darcy32

> SirenFNO-Darcy32 - OneScience 在 ModelScope 开源的数据集。本仓库整理 NeuralOperator Team 在 Zenodo 发布的 Darcy Flow Dataset 中 32×32 分辨率的训练、测试张量，用于 SirenFNO 二维 Darcy 渗流实验。原始文件来自 darcy32.tgz，本仓库只保留解压后的两份 .pt 文件，不修改其数值。

OneScience/SirenFNO-Darcy32 是 ModelScope 魔搭社区上的数据集，存储大小 188 MB，采用 other 许可。

- **Repository**: OneScience/SirenFNO-Darcy32
- **License**: other
- **Storage size**: 188 MB
- **Downloads**: 1
- **Stars**: 0
- **Last updated**: 2026-11-19

Source: https://www.modelscope.cn/datasets/OneScience/SirenFNO-Darcy32

---

# SirenFNO-Darcy32

## 数据集介绍

本仓库整理 [NeuralOperator Team 在 Zenodo 发布的 Darcy Flow Dataset](https://zenodo.org/records/12784353) 中 32×32 分辨率的训练、测试张量，用于 SirenFNO 二维 Darcy 渗流实验。原始文件来自 `darcy_32.tgz`，本仓库只保留解压后的两份 `.pt` 文件，不修改其数值。

## 数据集支持的任务

输入空间变化的渗透系数场 `x`，预测对应的压力解场 `y`。SirenFNO 实验使用训练文件前 1000 个样本及测试文件前 200 个样本，保持原始顺序。

## 数据格式

| 文件 | 字段 | shape | dtype | 说明 |
| --- | --- | --- | --- | --- |
| `darcy_train_32.pt` | `x`, `y` | 各 `[5000,32,32]` | `float32` | 原始训练池 |
| `darcy_test_32.pt` | `x`, `y` | 各 `[1000,32,32]` | `float32` | 原始测试池 |

原始文件 SHA256：训练 `f44a1151802eaf2eac65ddaded1ecf105bac019add0e3d277a3e9939cca1cd8b`；测试 `f9fbe30065908327f132425d760d573ed7ce29cb2a54cfdb3fcefd3d80d7ac83`。

## 数据集使用方式

```bash
modelscope download --dataset OneScience/SirenFNO-Darcy32 --local_dir ./data
```

```python
import torch
train = torch.load('data/darcy_train_32.pt', map_location='cpu', weights_only=False)
test = torch.load('data/darcy_test_32.pt', map_location='cpu', weights_only=False)
print(train['x'].shape, train['y'].shape, test['x'].shape, test['y'].shape)
```

## OneScience 官方信息

| 平台 | OneScience 主仓库 | Skills 仓库 |
| --- | --- | --- |
| Gitee | https://gitee.com/onescience-ai/onescience | https://gitee.com/onescience-ai/oneskills |
| GitHub | https://github.com/onescience-ai/OneScience | https://github.com/onescience-ai/oneskills |

## 引用与许可证

- 原始数据：[Darcy Flow Dataset](https://zenodo.org/records/12784353)，DOI `10.5281/zenodo.12784353`，由 NeuralOperator Team 发布。
- 关联论文：[SirenFNO](https://ijcai.org/proceedings/2026/0535.pdf)。
- Zenodo 记录未显示明确的数据再分发许可证，因此本整理仓库标为 `other`；使用时请注明原始来源并核对原始发布条件。本仓库不将模型代码许可视为数据许可。
