metadata
language: en
tags:
- weather-forecasting
- diffusion-models
- rectified-flow
- meteorology
- pytorch
- deep-learning
license: mit
datasets:
- meteolibre
MeteoLibre Rectified Flow Model
This is a repo with differents models used for doing weather forecasting :
- epoch_126_mtg_meteofrance_.safetensors (model with sat + ground station) : config is model_v0_mtg_meteofrance
Model Description
- Model type: Rectified Flow Diffusion Model
- Architecture: 3D U-Net with FiLM conditioning
- Input: Meteorological data patches (12 channels, 3D spatio-temporal)
- Output: Generated weather forecast data
- Training data: MeteoLibre meteorological dataset
- Language(s): Python
- License: MIT
Intended Use
This model is designed for:
- Weather pattern generation and forecasting
- Meteorological data augmentation
- Research in atmospheric science and weather prediction
- Educational purposes in machine learning for climate modeling
Training
The model was trained using:
- Framework: PyTorch with Hugging Face Accelerate
- Optimizer: Adam (lr=5e-4)
- Batch size: 64
- Epochs: 200
- Precision: Mixed precision (bf16)
- Distributed training: Multi-GPU support
Ethical Considerations
- Weather forecasting models should be used responsibly
- Consider environmental impact of computational requirements
- Validate predictions against ground truth data
- Not intended for critical decision-making without human oversight
Citation
If you use this model in your research, please cite:
@misc{meteolibre-rectified-flow,
title={MeteoLibre Rectified Flow Weather Forecasting Model},
author={MeteoLibre Development Team},
year={2025},
publisher={Hugging Face},
url={https://huggingface.co/meteolibre-dev/meteolibre-rectified-flow}
}
Contact
For questions or issues, please open an issue on the MeteoLibre GitHub repository.