Metadata-Version: 2.4
Name: keras_climate
Version: 0.1.1
Summary: Keras 3 framework for climate modeling and Earth observation
Project-URL: Homepage, https://github.com/anas-rz/keras-climate
Project-URL: Documentation, https://anas-rz.github.io/keras-climate/
Project-URL: Repository, https://github.com/anas-rz/keras-climate
Requires-Python: >=3.9
Description-Content-Type: text/markdown
Requires-Dist: keras>=3.0
Requires-Dist: numpy

# keras_climate

![Keras Climate](docs/banner.png)

A Keras 3 (TensorFlow / JAX / PyTorch backend) framework of models for
climate modeling and Earth observation, plus tooling to port pretrained
weights from the original reference implementations (almost always
PyTorch) onto the Keras equivalents.

**[Full documentation](https://anas-rz.github.io/keras-climate/)**

## Install

```bash
pip install keras-climate
```

## Quick start

```python
from keras_climate.remote_sensing import UNet, SegFormer
from keras_climate.forecasting import PatchTST
from keras_climate.weather import ConvLSTMNowcaster
from keras_climate.foundation import PrithviClassifier

model = UNet(input_shape=(256, 256, 4), num_classes=5)   # 4-band (e.g. RGB+NIR) input
model.summary()

forecaster = PatchTST(seq_len=336, pred_len=96, num_channels=7)
```

All builders return a plain `keras.Model`, so standard `model.compile(...)`
/ `model.fit(...)` / `model.save(...)` all work unchanged.

See the [documentation](https://anas-rz.github.io/keras-climate/) for the
full model roster, the pretrained-weight-porting guide, and which models
ship a ready-to-download real checkpoint.
