polish some code (#55583)

paddle/fluid/distributed/auto_parallel/spmd_rules/rules.h

@@ -132,7 +132,7 @@ REGISTER_SPMD_RULE(tanh_shrink, ElementwiseSPMDRule);
 REGISTER_SPMD_RULE(thresholded_relu, ElementwiseSPMDRule);
 REGISTER_SPMD_RULE(trunc, ElementwiseSPMDRule);
 
-// matmul rule
+// layer_norm rule
 REGISTER_SPMD_RULE(layer_norm, LayerNormSPMDRule);
 
 // replicated rule

python/paddle/distributed/auto_parallel/static/completion.py

@@ -938,7 +938,7 @@ class Completer:
         """Complete annotation for the partial annotated serial_main_program.
         Arguments:
             serial_main_program: partial annotated serial_main_program.
-        Returns:e
+        Returns:
             serial_main_program: completed annotated serial_main_program.
         """
 

python/paddle/distributed/auto_parallel/static/engine.py

@@ -757,7 +757,7 @@ class Engine:
 
     def _parallel(self, mode, all_ranks=False):
         # Parallelize program based on the planner's results
-        # For now, the completer has to be passed to the planner,
+        # For now, the completer has to be passed to the Parallelizer,
         # because we may use it to complete the annotation of the backward and update.
         parallelizer = Parallelizer(
             mode,

python/paddle/distributed/auto_parallel/static/planner_v2.py

@@ -40,7 +40,7 @@ class Planner:
     def __init__(self, mode, dist_context):
         self._mode = mode
         self._dist_context = dist_context
-        self._load = False
+        self._load = False  # load dist_attr from file
 
         # NOTE: [HighOrderGrad]. There are grad ops in forward phase, and it need
         # dependency of backward-forward ops in forward completion.