docs: updated quantile huber_delta docstring

IntelLabs · Oct 7, 2024 · b9d68e5 · b9d68e5
1 parent f78d209
commit b9d68e5
Showing 1 changed file with 4 additions and 0 deletions.
diff --git a/matsciml/models/losses.py b/matsciml/models/losses.py
@@ -90,6 +90,10 @@ def __init__(
         weight predicted vs. actual margins, as computed with ``loss_func``.
         The mean of the weighted loss is then returned.
 
+        In the case of ``loss_func='huber'``, the ``huber_delta`` argument specifies
+        the margin used to switch between MAE and MSE losses. This value
+        is applied globally (i.e. regardless of the quantile).
+
         Parameters
         ----------
         quantile_weights : dict[float, float]