Metadata-Version: 2.1
Name: eclipse-pytorch
Version: 0.0.5
Summary: A pytorch implementation of Eclipse
Home-page: https://github.com/tcapelle/eclipse_pytorch/tree/master/
Author: Thomas Capelle
Author-email: thomascapelle@gmail.com
License: Apache Software License 2.0
Keywords: pytorch nowcasting solar energy
Platform: UNKNOWN
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: Natural Language :: English
Classifier: Programming Language :: Python :: 3.8
Classifier: License :: OSI Approved :: Apache Software License
Requires-Python: >=3.8
Description-Content-Type: text/markdown
Requires-Dist: pip
Requires-Dist: packaging
Requires-Dist: torch (>1.6)
Requires-Dist: fastcore
Provides-Extra: dev

# Eclipse
> Implementing <a href='https://arxiv.org/pdf/2104.12419v1.pdf'>Paletta et al</a> in Pytorch


Most of the codebase comes from [Fiery](https://github.com/wayveai/fiery)

![Image](nbs/images/eclipse_diagram.png)

## Install

```bash
pip install eclipse_pytorch
```

## How to use

```python
import torch

from eclipse_pytorch.model import Eclipse
```

```python
eclipse = Eclipse(horizon=5)
```

let's simulte some input images:

```python
images = [torch.rand(2, 3, 128, 128) for _ in range(4)]
```

```python
preds = eclipse(images)
```

you get a dict with forecasted masks and irradiances:

```python
len(preds['masks']), preds['masks'][0].shape, preds['irradiances'].shape
```




    (6, torch.Size([2, 4, 128, 128]), torch.Size([2, 6]))



## Citation

```latex
@article{paletta2021eclipse,
  title     = {{ECLIPSE} : Envisioning Cloud Induced Perturbations in Solar Energy},
  author    = {Quentin Paletta and Anthony Hu and Guillaume Arbod and Joan Lasenby},
  year      = {2021},
  eprinttype = {arXiv},
  eprint    = {2104.12419}
}
```

## Contribute

This repo is made with [nbdev](https://github.com/fastai/nbdev), please read the documentation to contribute


