fix: metric ranges in the validation space not the normalized space (e…

…cmwf#116)

fix: metric ranges in the validation space, not the normalized space

CHANGELOG.md

@@ -17,6 +17,7 @@ Keep it human-readable, your future self will thank you!
     - Enable longer validation rollout than training
 ### Added
 - Included more loss functions and allowed configuration [#70](https://github.com/ecmwf/anemoi-training/pull/70)
+   - Fix that applies the metric_ranges in the post-processed variable space [#116](https://github.com/ecmwf/anemoi-training/pull/116)
 - Sub-hour datasets [#63](https://github.com/ecmwf/anemoi-training/pull/63)
 - Add synchronisation workflow [#92](https://github.com/ecmwf/anemoi-training/pull/92)
 

src/anemoi/training/train/forecaster.py

@@ -98,7 +98,7 @@ def __init__(
 
         variable_scaling = self.get_feature_weights(config, data_indices)
 
-        self.val_metric_ranges, _ = self.get_val_metric_ranges(config, data_indices)
+        _, self.val_metric_ranges = self.get_val_metric_ranges(config, data_indices)
 
         # Kwargs to pass to the loss function
         loss_kwargs = {"node_weights": self.node_weights}