diff --git a/paddle/fluid/distributed/auto_parallel/spmd_rules/rules.h b/paddle/fluid/distributed/auto_parallel/spmd_rules/rules.h
index 6070539f7bc40b89d1101a4dce0bd77acebcd148..c58333d0fb7011623f28449cdc06b459af029f64 100644
--- a/paddle/fluid/distributed/auto_parallel/spmd_rules/rules.h
+++ b/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
diff --git a/python/paddle/distributed/auto_parallel/static/completion.py b/python/paddle/distributed/auto_parallel/static/completion.py
index d8636153ccf1f72041525134abde3d4420aade17..1f5ccc94c3c3e453cdb4a5a1579dc67fb01e1d3e 100644
--- a/python/paddle/distributed/auto_parallel/static/completion.py
+++ b/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.
         """
 
diff --git a/python/paddle/distributed/auto_parallel/static/engine.py b/python/paddle/distributed/auto_parallel/static/engine.py
index ee3ab4ea208a716d31c61d2b9e2a7c4474850825..4c0c2e2e4732c9e9498c7b7bfb517f8aa15b51a3 100644
--- a/python/paddle/distributed/auto_parallel/static/engine.py
+++ b/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,
diff --git a/python/paddle/distributed/auto_parallel/static/planner_v2.py b/python/paddle/distributed/auto_parallel/static/planner_v2.py
index f0ac925371055b1cb5e408321a075725f7d31ee3..6e318fb56e15ad5eb968da80638dada0c31c6d80 100755
--- a/python/paddle/distributed/auto_parallel/static/planner_v2.py
+++ b/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.