Conversation
bd66376 to
2d51778
Compare
* Added flexibility for more variables. Added normalization statistics from graphcast as json files, refactored metrics, datasets and tests as well as lightning modules. The latter have to account for nans in the data when including sea-ice-fraction and others * Added code to zero out NaNs in ground truth values for loss calculation * Renamed stats to norm, changed variable output for label wrapper, applied changes to configs * Removed import that causes error * Replaced norm scheme in graphcast statsconfig * Fixed bug in load_module due to missing stats_cfg and changed behaviour of to_xarray routines to account for levels * Added softlinks recommended by geoarches doc to .gitignore and removed pushed softlinks * Addressed issues with paramaters interpolate_nans and warning_on_nans * Added tensordict utility to test predictions for NaNs * Added documentation to the normalization module, changed name of norm file and norm module according to reviews, adapted norm files * Changes loss_coffs to compute_loss_coeffs_args for clarity, added test for graphcast stats * Added dimension names of the dataset as arguments for the dataloaders, added comment to test of era5 loaders with graphcast norm scheme * Fixed Bug in dcpp test due to missing dimension names * Added more flexibility for data naming (either names have leading zeros or not) * Refactored code to use dimension indexers for selecting dimensions and converting between tensordict and xarray, modified tests * Fixed bug with multistep rollout due to missing next-states, added torch.int64 to allow dates after 2038 * Changed behaviour of dimension indexers, added description, adapted eval_multistep and test files * fixed format problems --------- Co-authored-by: Renu Singh <renu.singh@inria.fr>
2d51778 to
3255e23
Compare
8fd6795 to
c326000
Compare
c326000 to
49f4b02
Compare
dcee04d to
f1a69a1
Compare
f1a69a1 to
fce4efa
Compare
robert-DL
reviewed
Sep 1, 2025
| cmap="plasma", | ||
| origin="upper", | ||
| extent=[-180, 180, -90, 90], | ||
| transform=ccrs.PlateCarree(), |
Collaborator
There was a problem hiding this comment.
Robinson gives the curved projection, but matter of taste
robert-DL
reviewed
Sep 1, 2025
|
|
||
| spatial_datas = [] | ||
| for base_dir in base_dirs: | ||
| spatial_data = torch.load( |
Collaborator
There was a problem hiding this comment.
To be consistent, I would argue that deciding for a storage format, e.g. netcdf, seems desirable to me.
robert-DL
approved these changes
Sep 1, 2025
Collaborator
robert-DL
left a comment
There was a problem hiding this comment.
Just minor comments. If we agree on the changes, we can do them in a separate merge request
| fig = plt.figure(figsize=(4, 2)) | ||
| num_plots = len(spatial_datas) | ||
| fig, axes = plt.subplots( | ||
| 1, num_plots, figsize=(6 * num_plots, 5), subplot_kw={"projection": ccrs.PlateCarree()} |
Collaborator
There was a problem hiding this comment.
I think single plots are better, one can align them later in latex files
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
add spatial_mse to DeterministicMetrics:
eval_multistep.py supports