Saved in:
Bibliographic Details
Main Authors: Lehmann, Fanny, Ozdemir, Firat, Soja, Benedikt, Hoefler, Torsten, Mishra, Siddhartha, Schemm, Sebastian
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.19088
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916809101279232
author Lehmann, Fanny
Ozdemir, Firat
Soja, Benedikt
Hoefler, Torsten
Mishra, Siddhartha
Schemm, Sebastian
author_facet Lehmann, Fanny
Ozdemir, Firat
Soja, Benedikt
Hoefler, Torsten
Mishra, Siddhartha
Schemm, Sebastian
contents Recent advances in AI weather forecasting have led to the emergence of so-called "foundation models", typically defined by expensive pretraining and minimal fine-tuning for downstream tasks. However, in the natural sciences, a desirable foundation model should also encode meaningful statistical relationships between the underlying physical variables. This study evaluates the performance of the state-of-the-art Aurora foundation model in predicting hydrological variables, which were not considered during pretraining. We introduce a lightweight approach using shallow decoders trained on the latent representations of the pretrained model to predict these new variables. As a baseline, we compare this to fine-tuning the full model, which allows further optimization of the latent space while incorporating new variables into both inputs and outputs. The decoder-based approach requires 50% less training time and 35% less memory, while achieving strong accuracy across various hydrological variables and preserving desirable properties of the foundation model, such as autoregressive stability. Notably, decoder accuracy depends on the physical correlation between the new variables and those used during pretraining, indicating that Aurora's latent space captures meaningful physical relationships. In this sense, we argue that an important quality metric for foundation models in Earth sciences is their ability to be extended to new variables without a full fine-tuning. This provides a new perspective for making foundation models more accessible to communities with limited computational resources, while supporting broader adoption in Earth sciences.
format Preprint
id arxiv_https___arxiv_org_abs_2506_19088
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Finetuning a Weather Foundation Model with Lightweight Decoders for Unseen Physical Processes
Lehmann, Fanny
Ozdemir, Firat
Soja, Benedikt
Hoefler, Torsten
Mishra, Siddhartha
Schemm, Sebastian
Machine Learning
Recent advances in AI weather forecasting have led to the emergence of so-called "foundation models", typically defined by expensive pretraining and minimal fine-tuning for downstream tasks. However, in the natural sciences, a desirable foundation model should also encode meaningful statistical relationships between the underlying physical variables. This study evaluates the performance of the state-of-the-art Aurora foundation model in predicting hydrological variables, which were not considered during pretraining. We introduce a lightweight approach using shallow decoders trained on the latent representations of the pretrained model to predict these new variables. As a baseline, we compare this to fine-tuning the full model, which allows further optimization of the latent space while incorporating new variables into both inputs and outputs. The decoder-based approach requires 50% less training time and 35% less memory, while achieving strong accuracy across various hydrological variables and preserving desirable properties of the foundation model, such as autoregressive stability. Notably, decoder accuracy depends on the physical correlation between the new variables and those used during pretraining, indicating that Aurora's latent space captures meaningful physical relationships. In this sense, we argue that an important quality metric for foundation models in Earth sciences is their ability to be extended to new variables without a full fine-tuning. This provides a new perspective for making foundation models more accessible to communities with limited computational resources, while supporting broader adoption in Earth sciences.
title Finetuning a Weather Foundation Model with Lightweight Decoders for Unseen Physical Processes
topic Machine Learning
url https://arxiv.org/abs/2506.19088