Fix explicit np.array conversion (#6359)

* Remove np.array
* Add dtype=object
* Add changelog

changelog/6282.bugfix.md

@@ -0,0 +1 @@
+Fixed a deprication warning that pops up due to changes in numpy

rasa/nlu/classifiers/diet_classifier.py

@@ -679,14 +679,14 @@ def _create_model_data(
                         self._tag_ids_for_crf(example, tag_spec)
                     )
 
-        X_sparse_sequence = np.array(X_sparse_sequence)
-        X_sparse_sentence = np.array(X_sparse_sentence)
-        X_dense_sequence = np.array(X_dense_sequence)
-        X_dense_sentence = np.array(X_dense_sentence)
-        Y_sparse_sequence = np.array(Y_sparse_sequence)
-        Y_sparse_sentence = np.array(Y_sparse_sentence)
-        Y_dense_sequence = np.array(Y_dense_sequence)
-        Y_dense_sentence = np.array(Y_dense_sentence)
+        X_sparse_sequence = np.array(X_sparse_sequence, dtype=object)
+        X_sparse_sentence = np.array(X_sparse_sentence, dtype=object)
+        X_dense_sequence = np.array(X_dense_sequence, dtype=object)
+        X_dense_sentence = np.array(X_dense_sentence, dtype=object)
+        Y_sparse_sequence = np.array(Y_sparse_sequence, dtype=object)
+        Y_sparse_sentence = np.array(Y_sparse_sentence, dtype=object)
+        Y_dense_sequence = np.array(Y_dense_sequence, dtype=object)
+        Y_dense_sentence = np.array(Y_dense_sentence, dtype=object)
         label_ids = np.array(label_ids)
         tag_name_to_tag_ids = {
             tag_name: np.array(tag_ids)

rasa/utils/tensorflow/model_data.py

@@ -386,7 +386,7 @@ def _balanced_data(self, data: Data, batch_size: int, shuffle: bool) -> Data:
         final_data = defaultdict(list)
         for k, values in new_data.items():
             for v in values:
-                final_data[k].append(np.concatenate(np.array(v)))
+                final_data[k].append(np.concatenate(v))
 
         return final_data