tf.accumulate_n -> tf.add_n (#2815)

`tf.accumulate_n` has been deprecated, and from the source code, I see that `tf.accumulate_n` is just a wrapper of `tf.add_n` since TF 2.1 (tensorflow/tensorflow@292d309). See: https://www.tensorflow.org/api_docs/python/tf/math/accumulate_n https://www.tensorflow.org/api_docs/python/tf/math/add_n Signed-off-by: Jinzhe Zeng <[email protected]>
deepmodeling · Sep 14, 2023 · 58dd3e2 · 58dd3e2
1 parent a8b3123
commit 58dd3e2
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/deepmd/descriptor/se_a.py b/deepmd/descriptor/se_a.py
@@ -1054,8 +1054,8 @@ def _filter(
                         # add zero is meaningless; skip
                         rets.append(ret)
                     start_index += self.sel_a[type_i]
-                # faster to use accumulate_n than multiple add
-                xyz_scatter_1 = tf.accumulate_n(rets)
+                # faster to use add_n than multiple add
+                xyz_scatter_1 = tf.add_n(rets)
             else:
                 xyz_scatter_1 = self._filter_lower(
                     type_i,