https://github.com/dataymeric/ClimODE

<aside> 💡

Explore the notion project ! We are very proud that this project was mostly done asynchronously by leveraging the potential of collaboration tools !

This database is still a bit raw. However, I found it interesting to share it as is. I hope you’ll appreciate its insights and approach it with kindness. A good start is to explore the nested tasks (especially in the Code section) in the following database:

Tasks

Open each tasks and see what’s inside ;)

</aside>

About this project

The ClimODE project involves the reimplementation of a paper on climate forecasting using physics-informed neural ODEs. The project includes tasks such as understanding the paper, implementation, experimental protocol, conducting additional experiments, analyzing results, and preparing a presentation.

Drafted notes

Reimplementation of a ICLR 2024 paper: “**ClimODE: Climate Forecasting With Physics-informed Neural ODEs”**

• Compréhension de l'article
• Implémentation
• Protocole expérimental (adaptation ou non), expériences complémentaires menées et analyse des résultats
• Présentation (poster + oral)

Idée d’axe expérimental :

• utiliser le modèle entrainé par les auteurs
• Feature suplémentaire:
  • Script d’évaluation standardisé fourni par WeatherBench 2
  • Données plus récentes

Tasks