ENH replace loss module Gradient boosting (scikit-learn#26278)

Author: lorentzenchr

doc/common_pitfalls.rst

@@ -211,7 +211,7 @@ method is used during fitting and predicting::
     >>> from sklearn.model_selection import cross_val_score
     >>> scores = cross_val_score(pipeline, X, y)
     >>> print(f"Mean accuracy: {scores.mean():.2f}+/-{scores.std():.2f}")
-    Mean accuracy: 0.45+/-0.07
+    Mean accuracy: 0.46+/-0.07
 
 How to avoid data leakage
 -------------------------

doc/whats_new/v1.4.rst

@@ -150,6 +150,11 @@ Changelog
   :pr:`13649` by :user:`Samuel Ronsin <samronsin>`,
   initiated by :user:`Patrick O'Reilly <pat-oreilly>`.
 
+- |Efficiency| :class:`ensemble.GradientBoostingClassifier` is faster,
+  for binary and in particular for multiclass problems thanks to the private loss
+  function module.
+  :pr:`26278` by :user:`Christian Lorentzen <lorentzenchr>`.
+
 - |Efficiency| Improves runtime and memory usage for
   :class:`ensemble.GradientBoostingClassifier` and
   :class:`ensemble.GradientBoostingRegressor` when trained on sparse data.