Reformulate some comments in executor_manager and executor_group (apa…

…che#6740)

python/mxnet/executor_manager.py

@@ -30,7 +30,7 @@ def _split_input_slice(batch_size, work_load_list):
     Raises
     ------
     ValueError
-        If there are two many splits such that some slice can be empty.
+        In case of too many splits, leading to some empty slices.
     """
     total_work_load = sum(work_load_list)
     batch_num_list = [round(work_load * batch_size / total_work_load)
@@ -44,7 +44,7 @@ def _split_input_slice(batch_size, work_load_list):
         begin = int(min((end, batch_size)))
         end = int(min((begin + batch_num, batch_size)))
         if begin >= end:
-            raise ValueError('Too many slices such that some splits are empty')
+            raise ValueError('Too many slices. Some splits are empty.')
         slices.append(slice(begin, end))
     return slices
 

python/mxnet/module/executor_group.py

@@ -111,8 +111,8 @@ class DataParallelExecutorGroup(object):
     shared_group : DataParallelExecutorGroup
         Defaults to ``None``. This is used in bucketing. When not ``None``, it should be a executor
         group corresponding to a different bucket. In other words, it will correspond to a different
-        symbol but with the same set of parameters (e.g. unrolled RNNs with different lengths).
-        In this case, many memory will be shared.
+        symbol with the same set of parameters (e.g. unrolled RNNs with different lengths).
+        In this case the memory regions of the parameters will be shared.
     logger : Logger
         Default is `logging`.
     fixed_param_names: list of str