Jk/develop/gamma decay #998
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…evelop/fe_experiments
…evelop/fe_experiments
…evelop/fe_experiments
Jubeku
added
enhancement
config/default_config.yml
Outdated
| - "mse" | ||
| - 1.0 | ||
|
|
||
| timestep_weight: |
There was a problem hiding this comment.
is this backwards compatible?
| return (max_value - min_value) * np.cos(latitudes_radian) + min_value | ||
|
|
||
|
|
||
| def gamma_decay(forecast_steps, gamma): |
There was a problem hiding this comment.
maybe we can put this in a util file?
|
Wiki page for loss weighting options: https://gitlab.jsc.fz-juelich.de/esde/WeatherGenerator-private/-/wikis/home/Model-and-training-configuration/Loss-weighting |
|
Forecast fine-tuning experiments: As expected, gamma decay helps to improve scores in the early steps as they are weighted more during rollout training (compare red 8-step training to blue baseline). 
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
This PR enables to provide a weighting function to distinctly weight forecast steps in the loss calculations.
The implemented function is a gamma function with a scale factor. The name of the weighting function and the scale factor are provided in the
default_config.yaml.In principle other functions could be implemented within the same structure.
Issue Number
Closes #904
Checklist before asking for review
./scripts/actions.sh lint./scripts/actions.sh unit-test./scripts/actions.sh integration-testlaunch-slurm.py --time 60