Skip to content
/ kosmoss Public

Bootcamps for Deep Learning applied to a Weather Forecast UC: a) large dataset preprocessing, b) multi-node multi-GPU neural models training, and c) hyper-optimization

License

Notifications You must be signed in to change notification settings

alxyok/kosmoss

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

kosmoss

Content

What this is: a bootsrap to scaling mechanics.

What this is not:

  • A gentle introduction to Deep Learning or Machine Learning
  • An in-depth look at cutting-edge innovations in AI

Install

Prerequisites: valid CUDA install.

To install:

  • git clone this repo and cd into it
  • python -m pip install . to install this project as a library with its dependencies
  • python src/kosmoss/download.py to download data for both PyTorch + Lightning and TensorFlow + Keras content

One note though

  1. All of the material is based on the work of AI4sim team on a real-world use-case for ECMWF
  2. This is just an introduction, not a comprehensive, in-depth, hands-on material
  3. No low-level optimizations

About

Bootcamps for Deep Learning applied to a Weather Forecast UC: a) large dataset preprocessing, b) multi-node multi-GPU neural models training, and c) hyper-optimization

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published