microsoft / microsoft/aurora

Efficient Way to Pre-Download ERA5 inputs and Load them for Fine-Tuning

Open
#150 8 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
1k
Forks
174
PR merge metrics
No merged PRs in 30d

Description

I’m building on top of Aurora and adding custom decoders to predict additional variables. To do this, I still need to feed ERA5 data into Aurora to generate the latent representation before passing it to the custom decoders.

The documentation shows how to download ERA5 data for a specific day. However, what’s the best practice for pre-downloading all required ERA5 data so it can be efficiently loaded during model training, without becoming a bottleneck?

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the linked example_era5 documentation and trace the ERA5 loading path used to feed Aurora before custom decoders. Done means documenting a reproducible way to pre-download and load the required inputs efficiently during fine-tuning, including how to avoid a training bottleneck.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
data-engineering, machine-learning
Issue type
Documentation
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.