Minor doc clarification for reduce_sum return type

PiperOrigin-RevId: 191914398

diff --git a/tensorflow/python/ops/math_ops.py b/tensorflow/python/ops/math_ops.py
index b460ce5..01d670e 100644 (file)
--- a/tensorflow/python/ops/math_ops.py
+++ b/tensorflow/python/ops/math_ops.py
@@ -1402,10 +1402,11 @@ def reduce_sum(input_tensor,
     keep_dims: Deprecated alias for `keepdims`.
 
   Returns:
-    The reduced tensor.
+    The reduced tensor, of the same dtype as the input_tensor.
 
   @compatibility(numpy)
-  Equivalent to np.sum
+  Equivalent to np.sum appart the fact that numpy upcast uint8 and int32 to
+  int64 while tensorflow returns the same dtype as the input.
   @end_compatibility
   """
   keepdims = deprecation.deprecated_argument_lookup("keepdims", keepdims,