Use TruncatedNormal from Keras initializers (#24036)

Co-authored-by: Andrey Voynov <avoin@google.com>

src/transformers/modeling_tf_utils.py

@@ -3391,14 +3391,14 @@ class TFSequenceSummary(tf.keras.layers.Layer):
         return output
 
 
-def get_initializer(initializer_range: float = 0.02) -> tf.initializers.TruncatedNormal:
+def get_initializer(initializer_range: float = 0.02) -> tf.keras.initializers.TruncatedNormal:
     """
-    Creates a `tf.initializers.TruncatedNormal` with the given range.
+    Creates a `tf.keras.initializers.TruncatedNormal` with the given range.
 
     Args:
         initializer_range (*float*, defaults to 0.02): Standard deviation of the initializer range.
 
     Returns:
-        `tf.initializers.TruncatedNormal`: The truncated normal initializer.
+        `tf.keras.initializers.TruncatedNormal`: The truncated normal initializer.
     """
     return tf.keras.initializers.TruncatedNormal(stddev=initializer_range)